Is Treaty Interpretation an Art or a Science? International Law and Rational Decision Making

Abstract

Although treaty interpretation is undoubtedly an activity governed by international law, and by Articles 31–33 of the 1969 Vienna Convention on the Law of Treaties (VCLT) in particular, some commentators continue repeating the pre-Vienna adage that treaty interpretation is a matter of art and not science, the implication of which is that no understanding of a treaty provision can ever be explained rationally. As the present article argues, this idea of interpretation must be rejected. While, sometimes, an assumed meaning of a treaty cannot be justified based on international law simpliciter, many times it can still be explained based on the structural framework of Articles 31–33 of the VCLT. Consequently, any characterization of treaty interpretation in the abstract as either art or science is misplaced. Whether treaty interpretation is an art or a science remains a question of fact inextricably tied to the approach taken by each and every law-applying agent in particular cases.

1 Introduction

Whenever an issue of treaty interpretation causes dispute, attention will inevitably focus upon Articles 31–33 of the 1969 Vienna Convention on the Law of Treaties (VCLT).¹ This importance paid to Articles 31–33 in the practice of international law can be explained partly by the wide recognition of these articles as a reflection of customary international law.² It can also partly be explained by the second branch of the doctrine of inter-temporal law,³ which implies that in the interpretation of a treaty, law-applying agents have to comply, not with the rules of international law that possibly existed at the time of the conclusion of the treaty, but, rather, with the rules that exist at the time of interpretation.⁴

Taking a closer look at Articles 31–33, we find that generally they do not state the relevant international law in the form of interpretative directives – they do not give explicit instructions on how to arrive at a conclusion about the meaning of an interpreted treaty provision. The drafters of the VCLT have walked a delicate balance between the need to codify and clarify the law practised by international courts and tribunals and the wish to establish a set of norms that can be applied to treaties generally. This is why they have chosen a design that places primary emphasis on means of interpretation. Consequently, Article 31 stresses the importance of conventional language, adding that terms of a treaty must also be interpreted in their context and in light of the treaty’s object and purpose. Article 32 confirms that recourse may be had to supplementary means of interpretation, including the preparatory work of the treaty and the circumstances of its conclusion. This general outline of the Convention may explain why some commentators continue repeating the pre-Vienna adage that treaty interpretation is a matter of art and not science.⁵

As will gradually transpire in this article, the characterization of treaty interpretation as art is a good indicator of the fundamentally different views still existing among international lawyers on treaty interpretation matters and the proper or correct way to read Articles 31–33. This characterization suggests that although treaty interpretation is now governed by international law, no understanding of a treaty provision can ever be rationally explained.⁶ Certainly, this suggestion can be interpreted differently. It can be interpreted as a remark about the discovery of a meaning of a treaty.⁷ As such, it may be correct although rather trivial. Considering the inherently individualized nature of any process of discovery, and the countless number of factors that potentially may have an influence upon it, we cannot seriously expect to find a regular pattern systematically employed in each instance of discovery by a law-applying agent of an assumed meaning of a treaty. Consequently, reasons suggest that the characterization of treaty interpretation as art and not science should be read instead as a remark about the justification of an assumed meaning of a treaty. Interpreted in this way, I personally have great difficulties accepting it. As far as my experience of international law and legal practice is concerned, if rationality means that observers should be able to reconstruct an assumed meaning of a treaty as a conclusion inferred from sound premises according to the accepted rules of inference,⁸ then assumed meanings of treaties can indeed be explained in rational terms, and, what is more, they can be explained based on the structural framework of Articles 31–33 of the VCLT. Consequently, whether treaty interpretation is an art or a science cannot be determined in the abstract, based on the mere nature of the VCLT or the particular activity in question. It is a question of fact inextricably tied to the approach taken by each and every law-applying agent in particular cases. I will spend the remainder of this article arguing this proposition exactly.

2 The Structure of Articles 31–33

A necessary first step in my line of argument is to clarify the structural framework of Articles 31–33 of the VCLT. Two questions need to be answered:

• 1. What is the ultimate aim of the treaty interpretation process?

• 2. What is the role of the various means of interpretation for achieving this same aim?

Beginning with the first question, naturally, the ultimate aim of the treaty interpretation process, as described in the VCLT, is to establish the legally correct meaning of the interpreted treaty. By the legally correct meaning of a treaty, international lawyers generally understand the communicative intention of the treaty parties – that is to say, the meaning that the parties intended the treaty to express.⁹ This is the explanation to why, according to Article 31, paragraph 4, of the VCLT, an ordinary meaning shall be given to the terms of a treaty only in so far as it cannot be established that the treaty parties intended differently – that the terms should be given instead a special meaning. This is also the explanation to why, in establishing any such special meaning,¹⁰ law-applying agents shall take into account ‘any agreement relating to the treaty which was made between all the parties in connection with the conclusion of the treaty’;¹¹ ‘any instrument which was made by one or more parties in connection with the conclusion of the treaty and accepted by the other parties as an instrument related to the treaty’;¹² ‘any subsequent agreement between the parties regarding the interpretation of the treaty or the application of its provisions’;¹³ ‘any subsequent practice in the application of the treaty which establishes the agreement of the parties regarding its interpretation’;¹⁴ ‘any relevant rules of international law applicable in the relations between the parties’;¹⁵ the state or states of affairs that the parties to the treaty assumedly intended to attain by the application of the treaty;¹⁶ as well as the state or states of affairs that assumedly caused the parties to conclude it.¹⁷

The communicative intention of the treaty parties can only be assumed. Thus, the interpretation of a treaty is no different than the understanding of any verbal utterance produced by a person or group of persons, whether orally or in writing. As emphasized by modern linguistics (pragmatics), an utterance can be understood only on the assumption that whoever produced it acted rationally. That is to say, in expressing her communicative intention, the utterer acted in conformity with some particular standard or standards of communication. To facilitate reference, henceforth, I will refer to any such assumption made by an agent in the interpretation of an utterance as a communicative assumption.¹⁸

Naturally, different kinds of communicative assumptions may be relevant for the interpretation of utterances depending on such things as, for instance, the functional or social reason causing them. In the specific context of treaty interpretation, as can be seen from the practice of international courts and tribunals, law-applying agents operate on assumptions such as the following examples: that treaty parties have expressed their intention arranging so that the treaty conforms to the lexicon, grammar and pragmatic rules of the language used for every authenticated version of it;¹⁹ that treaty parties have expressed their intention arranging so that a consistent meaning can be conferred on all words and lexicalized phrases used in the interpreted treaty;²⁰ that treaty parties have expressed their intention arranging so that no norm expressed in the treaty logically contradicts any other;²¹ that treaty parties have expressed their intention arranging so that no part of the treaty comes out as redundant;²² that treaty parties have expressed their intentions arranging so that the application of the treaty results in the realization of its object and purpose;²³ that treaty parties have expressed their intention arranging so that the treaty does not derogate from any other international legal norm applicable in the relationship between them;²⁴ that treaty parties have expressed their intention arranging so that when the treaty expressly limits the scope of a generically defined class of referents it excludes all other referents belonging to this class;²⁵ that treaty parties have expressed their intention favouring the sovereign freedom of states;²⁶ that treaty parties have expressed their intention arranging so that the treaty corresponds to whatever can be inferred from the subsequent practice developed in its application, rather than whatever can be inferred from its preparatory work;²⁷ that treaty parties have expressed their intention arranging so that the treaty comes out as altogether logically consistent, rather than corresponding to whatever can be inferred from its preparatory work.²⁸

Turning now to the second question, the role of the various means of interpretation specified in the VCLT is obviously to permit assumptions such as those just stated. These assumptions are of two kinds. The first kind of assumption describes a relationship between the interpreted treaty and a particular means of interpretation. For example, in assuming that treaty parties have expressed their intention arranging so that the treaty conforms to the relevant lexicon, grammar and rules of pragmatics, law-applying agents make an assumption about the relationship between the interpreted treaty and conventional language. In assuming that treaty parties have expressed their intention so that no part of the treaty comes out as redundant, the law-applying agents make an assumption about the relationship between the interpreted treaty and the context. I will refer to assumptions of this kind as first-order communicative assumptions.²⁹

A second kind of assumption describes a relationship between two first-order communicative assumptions. For example, law-applying agents may assume that the treaty parties have expressed their intention arranging so that the treaty corresponds to whatever can be inferred from the subsequent practice developed in its application, rather than to whatever can be inferred from its preparatory work. Similarly, they may assume that treaty parties have expressed their intention arranging so that the treaty comes out as altogether logically consistent, rather than corresponding to whatever can be inferred from its preparatory work. I will refer to assumptions of this kind as second-order communicative assumptions.³⁰

As it appears, Articles 31–33 of the VCLT presuppose the existence of a series of first- and second-order communicative assumptions. If we can establish these assumptions, the information already provided in Articles 31–33 will allow a description of the substance of these articles in the form of interpretative directives or – as some would have it – proper rules of interpretation.³¹ Such rules of interpretation would then read something along the line of the following examples:

• Rule 1: If a treaty uses elements of conventional language (such as, for instance, words, grammatical structures, or pragmatic features), the treaty shall be understood in accordance with the rules of that language.

• Rule 2: If one of the two possible ordinary meanings of a treaty provision makes a part of the treaty redundant, whereas the other ordinary meaning does not, then the latter meaning shall be adopted.

• Rule 3: If one of the two possible ordinary meanings of a treaty provision helps attain the object and purpose of the treaty, whereas the other ordinary meaning does not, then the former meaning shall be adopted.

• Rule 4: If one of the two possible ordinary meanings of a treaty provision helps to favour the sovereign freedom of states, whereas the other ordinary meaning does not, then the former meaning shall be adopted.

• Rule 5: Rule 2 shall be applied prior to Rule 4, insofar as this does not leave the meaning of the treaty ambiguous or obscure or leads to a manifestly absurd or unreasonable result.

This list of examples helps emphasize the inconclusive nature of the law laid down in Articles 31–33 of the VCLT. If Articles 31–33 help law-applying agents resolve many issues of treaty interpretation, they obviously do not help them resolve all such issues. First, international law cannot always determine the extension of a means of interpretation relative to the particular issue confronted.³² For example, whereas, according to international law, law-applying agents shall understand a treaty in conformity with conventional language, international law does not tell those agents whether, in the interpretation of a particular treaty, conventional language shall be understood to mean the language applied at the time of its conclusion or the language applied at the time of interpretation. Second, international law cannot always determine the existence of the necessary relationship between the means of interpretation drawn upon and the interpreted treaty provision. For example, whereas, according to international law, if the ordinary meaning of a treaty is ambiguous, law-applying agents shall adopt the meaning that best helps attain the object and purpose of the treaty, international law cannot help those agents determine the instrumental relationship between the ordinary meanings of a treaty and its object and purpose. Third, international law cannot always determine the priority of two or more rules of interpretation. For example, whereas, according to international law, if a conflict occurs between Rules 2 and 4, the former shall normally have precedence, international law cannot help law-applying agents resolve a conflict between Rules 2 and 3.

Despite the existence of Articles 31–33 of the VCLT, to some extent, issues of interpretation still have to be resolved at the discretion of the law-applying agents themselves. The crucial question is whether this makes treaty interpretation an art and not science. As I will argue in the following section, the answer to this question inevitably depends on the approach taken by each and every law-applying agent in disposing of the discretion given to her. To establish this proposition, in sections 3–5 of the article, I will provide a series of examples taken from the practice of international courts and tribunals. As the examples go to show, when international judiciaries decide issues of interpretation that cannot be resolved on the basis of international law simpliciter – whether they concern the extension of a means of interpretation (section 3), the relationships between a means and an interpreted treaty (section 4), or the priority of the rules of interpretation (section 5) – typically, judiciaries still take great pains to explain their decisions in rational terms, using forms of reasoning firmly anchored in the structural framework of Articles 31–33 of the VCLT.

3 Explaining Assumptions about the Extension of a Means of Interpretation

A first-order communicative assumption is an assumption made by a law-applying agent about the relationship between the interpreted treaty and a particular means of interpretation. In determining the extension of a means of interpretation relative to a particular treaty and a particular issue of interpretation confronted, as noted in section 2 of this article, law-applying agents are often left with a certain scope of discretion. The following four examples will give an idea of how this discretion is typically used.

In the La Bretagne arbitration,³³ the resolution of a fishing dispute between France and Canada prompted the arbitration tribunal to engage with the ordinary meaning of the expression ‘fishery regulations’ used in a bilateral agreement concluded by the two states in 1972.³⁴ The dispute arose when, in January 1985, Canadian authorities rejected an application for a licence to fish in the Gulf of St. Lawrence submitted by the owner of a French trawler (La Bretagne) registered in St. Pierre and Miquelon. Canada defended this decision arguing, first, that Canadian authorities had acted pursuant to a long-standing policy of refusing to grant licences to all vessels equipped for filleting at sea, including those of Canadian nationality and, second, that this policy fell within the scope of the expression ‘fishery regulations’ in the sense of Article 6, paragraph 1, of the 1972 Agreement. The provision reads as follows:

Canadian fishery regulations shall be applied without discrimination in fact or in law to the French fishing vessels covered by Articles 3 and 4 [among others, French trawlers registered at St. Pierre et Miquelon], including regulations concerning the dimensions of vessels authorized to fish less than 12 miles from the Atlantic coast of Canada.

The Tribunal noted that the contents of fishery regulations generally had evolved to some extent since the adoption of the 1972 Agreement:

[W]hereas at the time of the conclusion of the Agreement, the fishery regulations in force in various States usually confined themselves to specifying forbidden fishing zones or closed seasons, permitted fishing gear and equipment, and the types, age and size of the species that could be caught, the scope of fishery regulations has since been enlarged; this applies to the regulations of both the Parties to the present case. Concern over the more efficient management of fish stocks has led to the introduction of other methods of supervising fishing efforts partly in the form of quotas for individual vessels within the total allowable catch (TAC) and partly in the form of fishing licences or permits for foreign vessels.³⁵

Interestingly, Article 31, paragraph 1, of the VCLT does not say whether the ordinary meaning of a treaty term should be determined based on the general usage of that term at the time of conclusion of the treaty or the usage at the time of interpretation. This did not prevent the Tribunal from concluding that just because the Canadian policy did not concern fishing directly it was not extraneous to the ordinary meaning of ‘fishery regulations’:

[T]he rules to which the expression [‘fishery regulations’] refers must not only be taken to be those setting technical standards for the physical conditions in which the fishing is carried on but also those requiring the completion of certain formalities prior to the performance of these activities.³⁶

This proposition presupposes the communicative assumption that the parties to the Agreement expressed their intention conforming to the usage of the English language in 1972. More specifically, it presupposes, first, that according to the lexicon of the English language in 1972 the term ‘fishery regulations’ could be used to stand for a class of administrative measures (rather than any specifically defined group of measures) and, second, that the parties to the 1972 Agreement, for the purpose of Article 6, used the term fishery regulations to stand for a class of administrative measures, which at some time during the life span of the treaty would probably undergo changes (rather than not). The explanation given by the Tribunal to substantiate its conclusion would seem to support this assumption. The Tribunal referred to Article 6, paragraph 3, which spoke of the promulgation of ‘new regulations applicable to these vessels’, thus showing beyond any doubt that Article 6, paragraph 1, did not have not have the effect of subjecting vessels only to the regulations in force at the time of the conclusion of the Agreement. It cited the previous practice of the parties to the 1972 Agreement and the circumstances of its conclusion:

[I]t was the 1972 Agreement which made the Canadian fishery regulations applicable to French vessels in the fishing zones established by Canada in 1964 and 1970: indeed, the Coastal Fisheries Protection Regulations of 17 July 1964 stipulated that Canadian fisheries legislation was not applicable to the sectors where these vessels operated in order to facilitate the conclusion of the negotiations then being conducted.³⁷

It finally recalled the unlimited temporal scope of the Agreement:

[A]s this expression [‘fishery regulations’] was embodied in an agreement concluded for an unlimited duration, it is hardly conceivable that the Parties would have sought to give it an invariable content.³⁸

In Guinea – Guinea-Bissau Maritime Delimitation,³⁹ a special agreement requested the Court of Arbitration to determine the legal effect of a series of documents annexed to the Franco-Portuguese Convention concluded in 1886 by France and Portugal.⁴⁰ The documents included 12 protocols containing the records of the negotiating sessions that once led to the adoption of the Convention. They included the entire texts of statements made in the course of those sessions as well as two maps. The disputing parties had all the reason in the world to be asking this question, since Articles 31 and 32 of the VCLT do not define either the concept of the text of a treaty or the concept of the preparatory work. Article 31, paragraph 2, merely restates the obvious: that for the purpose of the interpretation of a treaty, the context shall comprise the text of the treaty, ‘including its preamble and annexes’. The Court concluded that although all of the documents formally belonged to the annexes of the Convention, a distinction should be made between them. Consequently, according to the Court’s finding, the two maps were part of the text of the 1886 Convention; all of the other documents were part of the preparatory work.⁴¹ This finding is based on the communicative assumption that the parties to the 1886 Convention regarded the two maps, but not the other documents, as integral parts of the Convention.⁴² To support this assumption, the Court noted that in Articles 1 and 3 of the Convention, where the boundary separating French and Portuguese possessions is described, there were explicit references to the two maps,⁴³ whereas no such immediate role had been given by the treaty parties to the remainder of the ‘annexes’.⁴⁴

In the Case Concerning Soveriegnty over Pulau Ligitan and Pulay Sipadan (Indonesia/Malaysia),⁴⁵ the International Court of Justice (ICJ) had to decide, among other things, whether or not a Convention concluded by the Netherlands and Great Britain in 1891 established the sovereignty of either party over two small islands located off the north-east coast of Borneo.⁴⁶ To do so, it had to engage with various suggestions as to what might actually be the object and purpose of this Convention. As the Court confirmed, ‘the object and purpose of the 1891 Convention was the delimitation of boundaries between the parties’ possessions within the island of Borneo itself’.⁴⁷ Based on an assumption about the purposive intention of the treaty parties, obviously, this proposition begs the support of inferential evidence. This is why the Court took pains to show that a delimitation of boundaries within Borneo, but not beyond it, was exactly what the parties to the 1891 Convention had wished to achieve. The ICJ cited first of all the preamble of the Convention, which provided that the parties were ‘desirous of defining the boundaries between the Netherlands possessions in the Island of Borneo and the States in that island which are under British protection’.⁴⁸ It further found its interpretation supported by the general scheme of the 1891 Convention and the circumstance of its conclusion. As the Court explained:

Article I expressly provides that ‘[t]he boundary … shall start from 4 ° 10 ′ north latitude on the east coast of Borneo. Articles II and III then continue the description of the boundary line westward, with its endpoint on the west coast being fixed by Article III. Since difficulties had been encountered concerning the status of the island of Sebatik, which was located directly opposite the starting point of the boundary line and controlled access to the rivers, the parties incorporated an additional provision to settle this issue. The Court does not find anything in the Convention to suggest that that the parties intended to delimit the boundary between their possessions to the east of the islands of Borneo and Sebatik or to attribute sovereignty over any other islands.⁴⁹

In EC – Poultry,⁵⁰ the several issues raised by the Appellant (Brazil) urged the World Trade Organization’s Appellate Body to determine the relationship between an annex to the Marrakesh Protocol to the 1994 General Agreement on Tariffs and Trade (GATT) (referred to by the Appellate Body as ‘Schedule LXXX’) and a bilateral agreement (Oilseeds Agreement), which was signed by the European Communities and Brazil only four months before the conclusion of the Marrakesh Protocol.⁵¹ The appellant argued that Schedule LXXX either superseded or terminated the Oilseeds Agreement in the sense of Articles 30 and 59 of the VCLT, respectively. The Appellate Body explained that since it was Schedule LXXX that provided the basis for the dispute, and the Oilseeds Agreement was not part of the 1994 GATT, it had no reason to engage with this issue. The Appellate Body immediately added that, of course, the Oilseeds Agreement ‘may serve as a supplementary means of interpretation of Schedule LXXX pursuant to Article 32 of the Vienna Convention’.⁵² Since the Oilseeds Agreement is clearly not part of the preparatory work of the 1994 GATT, what the Appellate Body appears to be saying is that the Agreement forms part of the circumstances of the conclusion of Schedule LXXX. This proposition is based on the communicative assumption that the Oilseeds Agreement exerted an influence on the parties to Schedule LXXX and that they expressed their intention accordingly.⁵³ Some evidence is furnished by the Appellate Body to support this assumption when in characterizing the Oilseeds Agreement it referred to it as ‘part of the historical background of the concession of the European Communities for frozen poultry meat’.⁵⁴ To explain this characterization in more detail, the Appellate Body noted, first, ‘that the Oilseeds Agreement was negotiated within the framework of Article XXVIII of the GATT 1947 with the authorization of the contracting parties’.⁵⁵ Second, it recognized the fact ‘that both parties agree that the substance of the Oilseeds Agreement was the basis for the 15,000 tonne tariff-rate quota for frozen poultry meat that became a concession of the European Communities in the Uruguay Round set forth in Schedule LXXX’.⁵⁶

4 Explaining Assumptions about the Relationships between a Means and an Interpreted Treaty

A first-order communicative assumption entails that some particular kind of relationship exists between an interpreted treaty and a particular means of interpretation. In determining the existence of such a relationship, as noted in section 2 of this article, law-applying agents are often left with a certain scope of discretion. The following four examples will give an idea of how this discretion is typically used.

In La Grand Case,⁵⁷ Germany’s submissions prompted the ICJ to clarify the meaning of Article 41 of its own Statute.⁵⁸ According to Article 41, ‘[t]he Court shall have the power to indicate, if it considers that circumstances so require, any provisional measures which ought to be taken to preserve the respective rights of either party’. Based on the ordinary meaning and the object and purpose of the provision, the Court concluded ‘that the power to indicate provisional measures entails that provisional measures should be binding’.⁵⁹ It emphasized that it did not consider it necessary to also resort to the preparatory work of the ICJ Statute, but, nevertheless, it found it worthwhile to point out that the preparatory work did not preclude its conclusion.⁶⁰ This proposition assumes that the parties to the ICJ Statute expressed their intention arranging so that the Statute accords with whatever can be inferred from its preparatory work. More specifically, it assumes that preparatory work gives expression to an intention that orders decided under Article 41 are legally binding. Obviously, international law cannot justify this assumption. This explains why the Court proceeded the way it did. It cited the initial preliminary draft of the identically worded Statute of the Permanent Court of International Justice, which was prepared by the Committee of Jurists, following a text in French proposed by the Brazilian jurist Raul Fernandez. Translated into English, the text reads as follows:

In case the cause of the dispute should consist of certain acts already committed or about to be committed, the Court may, provisionally and with the least possible delay, order [‘la Court pourra ordonner’, in French] adequate protective measures to be taken, pending the final judgment of the Court.⁶¹

As the Court noted:

The Drafting Committee prepared a new version of this text [referred to as draft Article 2bis], to which two main amendments were made: on the one hand, the words ‘la Cour pourra ordonner’ (‘the Court may … order’) were replaced by ‘la Cour a le pouvoir d’indiquer’ (‘the Court shall have the power to suggest’), while, on the other, a second paragraph was added providing for notion to be given to the parties and to the Council of the ‘measures suggested’ [‘cette suggestion’] by the Court … The Committee of Jurists eventually adopted a draft Article 39, which amended the former Article 2bis only in its French version: in the second paragraph, the words ‘cette suggestion’ were replaced in French by the words ‘l’indiqation’.⁶²

The ICJ then recapitulated the consideration of draft Article 39 by the Sub-Committee of the Third Committee of the first Assembly of the League of Nations:

[A] number of amendments were considered. Raul Fernandez suggested again to use the word ‘ordonner’ in the French version. The Sub-Committee decided to stay with the words ‘indiquer’, the Chairman of the Sub-Committee observing that the Court lacked the means to execute its decisions. The language of the first paragraph of the English version was then made to conform to the French text: thus the word ‘suggest’ was replaced by ‘indicate’, and ‘should’ by ‘ought to’. However, in the second paragraph of the English version, the phrase ‘measures suggested’ remained unchanged.⁶³

The Court inferred from this entire drafting history:

The preparatory work of Article 41 shows that the preference given in the French to ‘indiquer’ over ‘ordonner’ was motivated by the consideration that the Court did not have the means to assure the execution of its decisions. However, the lack of means of execution and the lack of binding force are two different matters. Hence, the fact that the Court does not itself have the means to ensure the execution of orders made pursuant to Article 41 is not an argument against the binding nature of such orders.⁶⁴

In Territorial and Maritime Dispute (Nicaragua v. Colombia),⁶⁵ the ICJ had to determine whether by the conclusion of a Treaty and a Protocol in 1928 and 1930, respectively, the two disputing parties had established a general line of delimitation separating their respective maritime areas.⁶⁶ To support its position that they had, Colombia invoked a series of maps, produced by its government unilaterally, dating back to the time following immediately upon the conclusion of the two agreements. It pointed out, first, that on those maps, the 82nd meridian of longitude west had been depicted as the maritime boundary between Colombia and Nicaragua and, second, that Nicaragua had never lodged any protest against them.⁶⁷ The argument implies the existence of a subsequent practice establishing the agreement of the parties regarding the interpretation of the 1928 and 1930 Treaty and Protocol, in the sense of Article 31, paragraph 3(b), of the VCLT. The Court found no reason to approve with Colombia’s line of argument. As the Court explained:

An examination of these maps indicates that the dividing lines on them are drawn in such a way along the 82^nd meridian between the San Andrès Archipelago and Nicaragua that they could read either as identifying a general maritime delimitation between the two States or as only a limit between the archipelagos. Given the ambiguous nature of the dividing lines and the fact that these maps contained no explanatory legend, they cannot be deemed to prove that both Colombia and Nicaragua believed that the Treaty and Protocol had effected a general delimitation of their maritime spaces. Nicaragua’s failure to protest the maps does not therefore imply an acceptance of the 82^nd meridian as the maritime boundary.⁶⁸

It is important to understand the basis for this conclusion. Obviously, it does not assume that the subsequent practice helps to establish an agreement that the 82nd meridian forms a general line of delimitation, but neither does it assume that the subsequent practice helps to establish the agreement to the contrary. The conclusion of the Court, rather, builds on the communicative assumption that the subsequent practice of the parties did not establish any agreement whatsoever. Thus, if we require that it be rationally explained, the onus cannot be very exacting. The observation that all maps lacked an explanatory legend would seem sufficient.

In Guinea – Guinea-Bissau Maritime Delimitation,⁶⁹ a Court of Arbitration had to determine whether Article 1 of a Convention concluded in 1886 by France and Portugal established a maritime boundary between the respective possessions of those two states in West Africa.⁷⁰ Recognizing that in the interpretation of a treaty, Article 31, paragraph 3(a), urged law-applying agents to take into account ‘any subsequent agreement between the parties regarding the interpretation of the treaty’, the Court noted the negotiations that took place between Portugal and France on 8–10 September 1959. The negotiations concerned the delimitation of the territorial sea and continental shelf between Senegal (for whose foreign affairs France then remained responsible) and Portuguese Guinea. They resulted in an agreement concluded by an exchange of letters on 26 April 1960. Obviously, this agreement was not applicable as such to the relations of the two disputing parties. Yet, according to the Court, it helped to justify its conclusion that until 1978, when the dispute between Guinea and Guinea-Bissau arose, ‘the States signatories to the 1886 Convention and their successor States interpreted the text of the final paragraph of Article I of this instrument as not having established a maritime boundary’.⁷¹ The Court seemed greatly concerned to explain this inference. It noted the minutes of the negotiations of 1959, which stated:

The Portuguese delegation has expressed to the delegation of the French Republic and of the Community its wish, by reference to the Franco-Portuguese Convention signed in Paris on 12 May 1886, to consider as part of Portugal’s internal waters those waters situated in the peri meter defined in Article I in fine of the said Convention. It has been agreed that the delegation of the French Republic and of the Community would recommend to the governments in Paris and Dakar not to contest any such decision.⁷²

As the Court argued:

Portugal apparently was not seeking to extend its territorial waters beyond the limits it had itself established. If it had considered that Article I in fine delimited a maritime boundary, it would surely not, in 1959, have sought to have waters situated within the perimeter defined in 1886 recognized as internal waters.⁷³

This reasoning provides justification of the necessary communicative assumption, namely that the 1960 exchange of letters expressed the intention of the parties to the 1886 Convention not to establish a maritime boundary.

In the Case Concerning Border and Transborder Armed Actions (Nicaragua v. Honduras),⁷⁴ the ICJ had to engage with the meaning of two provisions in the 1948 American Treaty on Pacific Settlement (Pact of Bogotá), namely Article XXXI and XXXII.⁷⁵ The articles read as follows:

Article XXXI

In conformity with Article 36, paragraph 2, of the Statute of the International Court of Justice, the High Contracting Parties declare that they recognize, in relation to any other American State, the jurisdiction of the Court as compulsory ipso facto, without the necessity of any special agreement so long as the present Treaty is in force, in all disputes of a juridical nature that arise among them concerning:

• (a) The interpretation of a treaty;

• (b) Any question of international law;

• (c) The existence of any fact which, if established, would constitute the breach of an international obligation;

• (d) The nature or extent of the reparation to be made for the breach of an international obligation.

Article XXXII

When the conciliation procedure previously established in the present Treaty or by agreement of the parties does not lead to a solution, and the said parties have not agreed upon an arbitral procedure, either of them shall be entitled to have recourse to the International Court of Justice in the manner prescribed in Article 40 of the Statute thereof. The Court shall have compulsory jurisdiction in accordance with Article 36, paragraph 1, of the said Statute.

As Honduras argued, the ICJ could only be seized under Article XXXI if, in accord ance with Article XXXII, there had been earlier recourse to conciliation. According to Nicaragua, each of the two articles provide a basis for the jurisdiction of the ICJ, the one independently of the other. The Court would therefore have jurisdiction to settle a dispute under the provisions of Article XXXI in the cases covered by that Article, regardless of whether or not there had previously been recourse to conciliation. The Court found Nicaragua’s interpretation to be the more convincing. As it inferred, Honduras’s interpretation was contrary to the object and purpose of the American states conferred on Article XXXII: ‘[T]o reinforce their mutual commitments with regard to judicial settlement.’⁷⁶ This conclusion builds on an assumption about the instrumental relationship between each of the two alleged meanings of Article XXXI and its stated object and purpose. It assumes that the application of Article XXXI will better reinforce the mutual commitments of the American states, if the meaning suggested by Nicaragua is preferred over the meaning suggested by Honduras. Typically, justifying an assumption of this kind is a tricky thing since, analytically speaking, it requires an assessment of the future effect of the application of the interpreted treaty.⁷⁷ The ICJ, however, faced a somewhat easier task. As it explained:

Honduras’s interpretation would … imply that the commitment, at first sight firm and unconditional, set forth in Article XXXI would, in fact, be emptied of all content if, for any reason, the dispute were not subjected to prior conciliation.⁷⁸

Arguably, if it is often difficult to explain why one meaning should be preferred over another, when assumedly both will result in the realization of the object and purpose of the interpreted treaty to some extent, the task is significantly easier when it is found that the one meaning will rule out its realization altogether. As in Border and Transborder Armed Actions, it can often be determined based merely on the text of the interpreted treaty.

5 Explaining the Priority of Different Rules of Interpretation

A second-order communicative assumption is an assumption made by a law-applying agent about the relationship between two first-order communicative assumptions. To some extent, Articles 31–33 determine the relationship between those first-order assumptions that draw upon the primary means of interpretation listed in Article 31 and those that draw upon the supplementary means of interpretation (Article 32).⁷⁹ Consequently, according to Article 32, if the application of Article 31 leaves the meaning of an interpreted treaty ambiguous or obscure, or leads to a result that is manifestly absurd or unreasonable, then whatever assumptions can be made based on Article 32 shall have priority over those based on Article 31. Inversely, if the application of Article 31 does not leave the meaning of an interpreted treaty ambiguous or obscure, and it does not lead to a result that is manifestly absurd or unreasonable, then whatever assumptions can be made based on Article 31 shall have priority over those based on Article 31. Articles 31–33, on the other hand, do not help determine when the application of Article 31 leads to a result that is manifestly absurd or unreasonable. Similarly, they do not help determine the relationship between the different communicative assumptions that can often be made based on Article 31 and the context and the object and purpose of a treaty. Neither do Articles 31–33 help determine the relationship between two or more communicative assumptions based on the supplementary means of interpretation. In other words, once again, international law leaves to law-applying agents a certain scope of discretion. The following examples will give an idea of how this discretion is typically used.

In Bosnia Genocide (Merits),⁸⁰ the ICJ had to determine whether states parties to the Genocide Convention, by virtue of the Convention, were under an obligation not to commit genocide themselves.⁸¹ As the Court readily admitted, such an obligation was not expressly imposed by the actual terms of the Convention. Still, as it concluded, ‘taking into account the established purpose of the Convention, the effect of Article I is to prohibit States from themselves committing genocide’.⁸² Article I of the Genocide Convention, it may be recalled, imposes upon parties to the Convention the obligation to prevent and to punish all acts of genocide. To explain its conclusion, the Court invoked among other things the following line of argument:

The expressly stated obligation to prevent the commission of acts of genocide … requires the State Parties, inter alia, to employ the means at their disposal … to prevent persons or groups not directly under their authority from committing an act of genocide or any of the other acts mentioned in Article III.⁸³

As recognized by the Court itself, this interpretation goes beyond the ordinary meaning of Article I. Thus, to support it, a mere reference to the purpose of Article I is not sufficient. It has to be established also that the ordinary meaning of Article I amounts to a ‘manifestly absurd or unreasonable’ result, in the sense of Article 32 of the VCLT. Stated in terms of the particular issue of interpretation, it has to be shown that, actually, very strong reasons support the following second-order communicative assumption: ‘Parties to the Genocide Convention expressed their intentions arranging so that Article I contributes to the realization of its object and purpose, rather than conforms to the lexicon, grammar, and pragmatic rules of the English language.’ International law fails to provide these reasons, and this would seem to be why the ICJ resorts to the following explanation:

It would be paradoxical if States were thus under an obligation to prevent, so far as within their power, commission of genocide by persons over whom they have a certain influence, but were not forbidden to commit such acts through their own organs, or persons over whom they have such firm control that their conduct is attributable to the State concerned under international law. In short, the obligation to prevent genocide necessarily implies the prohibition of the commission of genocide.⁸⁴

Arguably, this explanation, however terse, is exceptionally convincing.

In Soering v. the United Kingdom,⁸⁵ the European Court of Human Rights (ECtHR) had to engage with the suggestion made by Amnesty International that the death penalty should then (in 1989) be considered an inhuman and degrading punishment in the sense of Article 3 of the European Convention for the Protection of Human Rights and Fundamental Freedoms (ECHR). As the Court noted:

‘[T]he Convention is a living instrument … which must be interpreted in the light of present-day conditions’; and [as a consequence of this], in assessing whether a given treatment or punishment is to be regarded as inhuman or degrading for the purposes of Article 3 (art. 3), the Court cannot but be influenced by the developments and commonly accepted standards in the penal policy of the member States of the Council of Europe in this field’. De facto the death penalty no longer exists in time of peace in the Contracting States to the Convention. In the few Contracting States which retain the death penalty in law for some peacetime offences, death sentences, if ever imposed, are nowadays not carried out. This ‘virtual consensus in Western European legal systems that the death penalty is, under current circumstances, no longer consistent with regional standards of justice’, to use the words of Amnesty International, is reflected in Protocol No. 6 (P6) to the Convention, which provides for the abolition of the death penalty in time of peace.⁸⁶

Prima facie, as the Court seemed fully willing to admit, this development could in fact be seen to support the wide interpretation of Article 3 that Amnesty International had suggested. However, it was quick to add:

The Convention is to be read as a whole and Article 3 (art. 3) should therefore be construed in harmony with the provisions of Article 2 (art. 2) (see, mutatis mutandis, the Klass and Others judgment of 6 September 1978, Series A no. 28, p. 31, § 68). On this basis Article 3 (art. 3) evidently cannot have been intended by the drafters of the Convention to include a general prohibition of the death penalty since that would nullify the clear wording of Article 2 § 1 (art. 2-1).⁸⁷

If we are to analyse this reasoning, the ECtHR was obviously facing a conflict between two different first-order communicative assumptions. On the one hand, there is the assumption that the parties to the ECHR expressed their intention arranging so that the meaning of Article 3 continuously corresponds to whatever can be inferred from the subsequent practice developed in the application of this provision – in this case, the national penal policy of the European states. On the other hand, there is the assumption that the parties to the Convention expressed their intention arranging so that Article 3 comes out as being logically consistent with Article 2. Resolving this conflict, the Court obviously decided that the latter assumption should be preferred over the former. This decision builds on a second-order communicative assumption. It builds on the assumption that the parties to the ECHR expressed their intention arranging so that the treaty comes out as altogether logically consistent, rather than corresponding to whatever can be inferred from its preparatory work.

This assumption requires support that can only be found beyond international law simpliciter. There are two distinct ways to provide this support. The ECtHR can choose to try to either reinforce the strength of the preferred first-order communicative assumption or undermine the strength of the conflicting assumption. As shown by the following passage, the Court obviously opted for the latter alternative:

Protocol No. 6 (P6), as a subsequent written agreement, shows that the intention of the Contracting Parties as recently as 1983 was to adopt the normal method of amendment of the text in order to introduce a new obligation to abolish capital punishment in time of peace and, what is more, to do so by an optional instrument allowing each State to choose the moment when to undertake such an engagement. In these conditions, notwithstanding the special character of the Convention (see paragraph 87 above), Article 3 (art. 3) cannot be interpreted as generally prohibiting the death penalty.⁸⁸

Protocol no. 6 adds to the subsequent practice already considered – the national penal policy. It shows the overall practice to be less consistent than Amnesty International would otherwise have it.

In the Case of James and Others,⁸⁹ the ECtHR found itself faced with a very similar situation. The applicants in this case argued that the respondent (the United Kingdom), by a compulsory transfer of their property and by depriving them of their entitlement to proper compensation, had acted in violation of Article 1 of Protocol no. 1 to the ECHR. The relevant part of this Article reads as follows: ‘Every natural or legal person is entitled to the peaceful enjoyment of his possessions. No one shall be deprived of his possessions except in the public interest and subject to the conditions provided for by law and by the general principles of international law.’

One of the issues of interpretation brought before the Court was whether, as the applicant argued, the reference in the second sentence of Article 1 to ‘the general principles of international law’ implied that the international law requirement of prompt, adequate and effective compensation for the expropriation of property of foreigners applied also to nationals. The applicants contended that it did. According to them, to treat the general principles of international law as inapplic able to a taking by a state of the property of its own nationals would be tantamount to permitting differentiation on the ground of nationality. This, they said, would be incompatible with the expression: ‘no one’, which opened the second sentence of Article 1, and with Article 1 of the Convention, which by virtue of Article 5 of Protocol no. 1 obliged the respondent to secure to everyone within its jurisdiction all rights laid down in this Protocol.⁹⁰ The Court did not find this interpretation convincing. In its opinion, ‘the general principles of international law’ were not applicable to a taking by the state of the property of its own nationals, for the following reason:

Article 1 (P1-1) expressly provides that deprivation of property must be effected ‘in the public interest’: since such a requirement has always been included amongst the general principles of international law, this express provision would itself have been superfluous if Article 1 (P1-1) had had the effect of rendering those principles applicable to nationals as well as to non-nationals.⁹¹

Just as in Soering, the situation presents a conflict between two first-order communicative assumptions. On the one hand, there is the assumption implied by the argument of the applicants that the parties expressed their intention arranging so that Article 1 of Protocol no. 1 comes out as logically consistent with Article 5 of this same Protocol and with Article 1 of the Convention. On the other hand, there is the assumption implied by the reasoning of the Court that the parties expressed their intention arranging so that no part of Protocol no. 1 comes out as redundant. The interpretation of the Court implies a preference of the latter assumption over the former. It implies the following second-order communicative assumption: the parties to Protocol no. 1 expressed their intention arranging so that Protocol no. 1 comes out as altogether logically consistent, rather than avoiding within this Protocol all redundant expressions. Once again, as in Soering, the ECtHR, in trying to support this assumption, resorted to the strategy of undermining the strength of the first-order communicative assumption that it had chosen not to prefer:

As to Article 1 (art. 1) of the Convention, it is true that under most provisions of the Convention and its Protocols nationals and non-nationals enjoy the same protection but this does not exclude exceptions as far as this may be indicated in a particular text (see, for example, Articles 4 para. 3 (b), 5 para. 1 (f) and 16 of the Convention, Articles 3 and 4 of Protocol No. 4) (art. 4-3-b, art. 5-1-f, art. 16, P4-3, P4-4).⁹²

6 Conclusions

In order to come to a decision on the correct understanding of a disputed treaty provision, law-applying agents often have to decide issues of treaty interpretation that cannot be resolved on the basis of international law simpliciter. As noted in section 2 of this article, such issues may concern the extension of a means of interpretation relative to a particular treaty and a particular issue of interpretation. They may concern the existence of a relationship between a means of interpretation and a particular treaty provision or, again, they may concern the priority of two rules of interpretation. In all cases – section 2 made this very clear – the rationality of the decision of the agent inevitably turns on the justification of a communicative assumption. As shown by the examples given in sections 3–5, international courts and tribunals – owing to their particular entrusted task and their great authority as a material source of international law – often take great pains to provide this justification.

This is the simple explanation to why I object to the suggestion that treaty interpretation is a matter of art and not science. For me – and I like to think that this article shows me right – treaty interpretation is not necessarily either. If ever treaty interpretation is an art and not a science, consequently, as I insist, this is not because of the mere nature of the particular activity in question or because of any inherent shortcoming of Articles 31–33 of the VCLT. Rather, it is because Articles 31–33 leave to law-applying agents a certain scope of discretion and because, ultimately, law-applying agents have no obligation to approach this scope of discretion in any particular way. They may go down the track of rational decision making or they may not. Obviously, while not all law-applying agents find the idea of rational decision making appealing, if this article is correct, the claim that treaty interpretation is a matter of art and not science cannot serve as a valid excuse for choosing the latter alternative.