Referencing Patterns at the International Criminal Court

A largely neglected area of study in international law has been referencing patterns by international courts and tribunals. This article assesses referencing data collected from the International Criminal Court’s (ICC) records issued in the Uganda and Central African Republic situations. The data is generally restricted to ‘persuasive citations’ – those references that the ICC’s various chambers have used to help decide a point of law. Covering over 500 records, this study addresses, among other things, the frequency with which the ICC cites its own judgments, the nature of external sources cited, how referencing changes over time and how often individual judges cite their own decisions. The data may prove useful to the ICC itself, advocates who appear before it and scholars of international law.


Introduction
The extent to which international courts and tribunals use prior judgments as a basis for decision making has been a subject of increasing scholarly interest. 1 Still a relatively unexplored area of study, 2 to date only qualitative research has been conducted. 3 Qualitative research has its advantages, but it is unable to capture patterns that are only discernible by examining large quantities of citations. This article aims to partially address that lacuna through a quantitative study of the use of precedent by the International Criminal Court (ICC). 4 Counting citations to reveal patterns -often called 'citation analysis' -can answer questions that qualitative work cannot, such as: How frequently does the Court reference external sources of law and how often does it turn inward to its own decisions? Do the different divisions of the ICC -the Pre-Trial, Appeals and Trial Divisions -differ in their referencing rates and the sources that they cite? How do citation rates change over time? Which external courts are cited most frequently? How often do judges cite their own decisions? In answering these questions, this article covers the ICC decisions issued in the Uganda and Central African Republic (CAR) matters. 5 While the data collected encompasses a wide variety of sources to which the Court cites, the article focuses primarily on the ICC's use of its own previous decisions. In this study, 574 orders and decisions were analysed -slightly over 17 per cent of the ICC's total of 3,374, as of 2 May 2014, the final date on which data was collected.
This study did not use software to count citations (also called 'references'). Software is unable to discern whether a court is citing to a source to (i) support a point of law or (ii) merely indicate procedural history or cite the prosecution's or defence's submissions. In this article, the former is counted, while the latter is not. Recitations of judgments are not 'precedent' and, thus, counting them would undermine the usefulness of the data. Software is also not capable of determining, for instance, whether a citation is being used as a persuasive precedent or is being distinguished. In this study, each footnote in the 574 court records was individually scrutinized to determine how the court was using the sources in that footnote.
From the data gathered for this article, some generalizations about the ICC's referencing can be made. Referencing rates largely depend on, and reflect, the progress of a case. At the commencement of a case, the number of court records issued and the sources of law cited at the Pre-Trial Chambers is high. As the proceedings shift to the trial stage (as in the CAR case, which has a trial in progress), the burden to issue records, and, thus, an increase in citations, swings to the Trial Chamber. Overall, referencing rates of the combined chambers (but not necessarily between individual chambers) were comparable between the Uganda and CAR matters. The combined chambers adjudicating the Uganda case averaged slightly over nine references per court record, while those hearing the CAR case averaged nearly seven references per record. 6 4 The term 'precedent' as used in the context of international law can be ambiguous and confusing. See, e.g., Miller, ' An International Jurisprudence? The Operation of "Precedent" Across International Tribunals', 5(3) Leiden Journal International Law (2002) 483, at 488. While Miller prefers the term 'refer', this article uses 'precedent' when appropriate -for instance, to distinguish 'persuasive precedent' from 'binding precedent'. 5 A second investigation in the Central African Republic (CAR) was opened on 24 Sept. 2014 with respect to crimes allegedly committed since 2012. 6 These numbers exclude court records that cited no sources. The figures mean little without context. Courts cite sources of law for a myriad of reasons, and the amount per decision depends largely on the number and nature of the legal issues at hand as well as the availability of relevant sources of law. Some ICC decisions subjected to analysis extended to nearly 200 pages and contained long strings of citations, while others were less than five pages and contained one or two citations.
The Appeals and Pre-Trial Chambers generally refer at a higher rate than the Trial Chamber, although only slightly. 7 All three chambers handle complicated legal issues that require support through references, but it is not surprising that the Trial Chamber would address fewer such issues. The Pre-Trial Division, of which the Pre-Trial Chambers are a part, was established in part to relieve the Trial Division of at least some of the burden of motions that had prolonged trials at the International Criminal Tribunal for the Former Yugoslavia. 8 Free to focus on its primary duty -to ensure that a trial is fair and expeditious and is conducted with full respect for the rights of the accused and due regard for the protection of victims and witnesses 9 -the Trial Chamber could be expected to handle fewer disputes over points of law requiring extensive referencing. In the CAR case, the Pre-Trial Chambers averaged 9.6 references per record, the Trial Chamber 5.2 and the Appeals Chamber 5.6; at the Pre-Trial Chamber hearing the Uganda case, there were 8.1 references per record, while at the Appeals Chamber, 10.3.
Case law dominates the sources of law used by the ICC. The use of persuasive cases as a percentage of all sources of law 10 ranged from a low of 70 per cent at the Pre-Trial Chambers in the CAR matter to a high of 92 per cent at the Appeals Chamber in that matter. The second most common source of law was treatises, journal articles and dictionaries 11 (combined), which was 12 per cent at its highest rate. Among the case law referenced, the number of prior decisions from the ICC chambers themselves significantly overshadowed those from all of the other courts.
ICC judges not only cite precedent from other ICC chambers, but they also cite their own prior decisions. Although technically acceptable, self-citing, particularly when it is unusually high, raises a concern that judges are basing their legal opinions upon little more than their own prior opinions. 'Self-citing', as it is used in this article, does not mean a chamber merely citing itself but, instead, refers to a judge citing his or her own previous decision. Self-citing in the CAR matter -the only matter in which self-citing was analysed -was highest in the Appeals Chamber, where it was over three times the average in the Trial Chamber and over ten times the average in the Pre-Trial Chambers.
The second part of this article explains the use of precedent by international courts and tribunals, the sources of law -both mandatory and permissive -used by the ICC and the referencing of external case law. Turning to the data in this study, the third 7  2000) and travaux preparatoires. These sources were excluded because, as at least the first three are so clearly mandatory sources that must be initially consulted by the Court, they are unhelpful in illuminating the Court's exercise of discretion to choose between sources. 11 Although dictionaries are not a source of law, the data include dictionaries because the ICC judges use them as a source for interpreting words, and, thus, they can have a meaningful impact on the ICC's analyses.
part describes the scope of the data and methodology of collection. The fourth part of the article presents and analyses the data, while the fifth part concludes with the potential usefulness of the data and raises some areas for further study.

Referencing by International Courts and Tribunals A General Referencing Standards
The standards by which international courts and tribunals reference sources of law provide the crucial context in which this article's data should be interpreted. Generally, international courts and tribunals do not use stare decisis, the doctrine requiring judges to follow previous similar decisions. 12 Rather, a particular decision binds only the parties before the court. 13 As a result, prior decisions are of diminished importance at the international level, in contrast to the national level where precedent is binding on lower courts in common law countries such as the USA and England. An early sign that prior rulings would not be binding on international tribunals was given by the Permanent Court of International Justice (PCIJ), which was inaugurated in 1922. The creators of the PCIJ agreed that the PCIJ's decisions should merely state law and not create law. 14 The Statute of the PCIJ provided that judicial decisions were only a 'subsidiary means for the determination of rules of law'. 15 The International Court of Justice (ICJ) -the successor of the PCIJ -similarly confirmed that it is not required to follow its precedent, though it frequently cites its own decisions to ensure consistency in decision making. 16 In fact, the ICJ considers its previous decisions 'authoritative '. 17 The application of stare decisis in international court systems with appellate organs is less uniform. At the International Criminal Tribunals for the Former Yugoslavia (ICTY) and Rwanda (ICTR), 18 rationales from the appellate authorities are binding, not only in the case at hand but also in subsequent subordinate chambers' decisions. 19 Similarly, the European Court of Human Rights (ECtHR) relies heavily on prior decisions, at times referring to them as 'precedent'. 20  Body of the World Trade Organization do not officially bind the lower Dispute Settlement Panel, 21 'de facto norms of stare decisis' nonetheless apply. 22 The European Court of Justice, on the other hand, has shown little reluctance to distinguish, and, in some cases, explicitly overrule, precedent. 23 The key point for the purposes of this article is that at the ICC's founding in 1998 the standard that at single-level international courts precedent is treated with respect and deference but is not binding had become somewhat established by the PCIJ and ICJ, while at multi-level courts the treatment of precedent remained inconsistent.

B Referencing Rules at the ICC
Article 21 of the ICC's founding statute -the Rome Statute -is unique among international tribunals. 24 In contrast to the charters, statutes, laws and agreements of most other prominent international tribunals, 25 Article 21 specifically sets out the applicable law for the Court. 26 Additionally, unlike Article 38 of the Statute of the ICJ, the Rome Statute lays out a clear hierarchy of sources of law to be applied. 27 Under this hierarchy, the Court must first apply the Rome Statute itself. 28 If the Statute does not address the issue at hand, and the issue is related to genocide, crimes against humanity or war crimes, the Court must then turn to the Elements of Crimes, a document adopted by the ICC Assembly of States Parties that elaborates on the crimes described in the Rome Statute. 29 When relevant, the Court must also look to its Rules of Procedure and Evidence. 30 If the Rome Statute, the Elements of Crimes and the Rules of Procedure and Evidence do not adequately address an issue, Article 21(1)(b) requires the Court to then, where appropriate, apply applicable treaties and the principles and rules of international law. 31 If still not resolved, the Court must turn to a third category of law, namely the 'general principles of law derived by the Court from national laws of legal systems of the world including, as appropriate, the national laws of States 21 Guillaume, supra note 14, at 12. 22 Lupu and Voeten, supra note 3, at 3. 23  that would normally exercise jurisdiction over the crime'. 32 The Court must apply the sources of law heretofore mentioned. In contrast, the Court may, but need not, apply the principles and rules of law as interpreted in its own decisions. 33 There is no hierarchy among the decisions of the ICC's three divisions. 34 In practice, the Court did not wait long to use its decisions as precedent, and the use of precedent thus far has not indicated that the Appeals Chamber rulings are superior to those of other chambers. 35 Finally, the application and interpretation of all sources of law by the ICC must be consistent with internationally recognized human rights and cannot make adverse distinctions based on race, gender, language, wealth, age, colour, religion or belief, political or other opinion, national, ethnic or social origin, birth or other status. 36

C Referencing External Judicial Decisions
Citing by international courts and tribunals to external judicial decisions -that is, those from other courts -while widespread, varies in frequency and is diverse in form and content. 37 External decisions are used for a wide range of purposes, from assisting in the interpretation of procedural issues, to providing specific rules of law, to supplying general legal principles. 38 ICJ decisions (particularly on matters of general public international law) and the decisions of international and regional human rights courts (especially on due process issues) are given great deference and considered 'highly persuasive'. 39 External decisions may be persuasive, but they clearly have no binding force 40 and are considered only when there are no useful precedents from the courts' own jurisprudence. 41 Although the Rome Statute expressly permits the Court to utilize its own prior rulings in decision making, it does not expressly address the use of external judicial decisions. 42 While the ICC has referred to decisions of the ICTY and ICTR, as well as the 32 Rome Statute, supra note 9, Art. 21(1)(c ICJ and ECtHR, on substantive law, it has not done so on procedural issues because its rules concerning victims and the roles of the prosecutor and judges are different from the rules of ad hoc tribunals. 43 In Prosecutor v. Thomas Lubanga Dyilo,44 for instance, the ICC Trial Chamber relied extensively on decisions of other international criminal courts and tribunals such as the ICTY and the Special Court for Sierra Leone (SCSL) and also cited the decisions of the ECtHR. 45 ICC Pre-Trial Chamber I has used ICTY rulings to help it determine the meaning of 'international armed conflict', 'armed conflict not of an international character' and the relationship required between individual criminal conduct and the hostilities occurring. 46 In fact, references to ICTY and ICTR decisions are common. 47 Reportedly, participants in the proceedings have 'constantly' referred to the ad hoc tribunals in their submissions to the Court even though their jurisprudence is not applicable law under Article 21. 48 Perhaps in response, the ICC's Pre-Trial Chamber II announced that 'the law and practice of the ad hoc tribunals, which the Prosecutor refers to, cannot per se form a sufficient basis for importing into the Court's procedural framework remedies other than those enshrined in the Statute'. 49 National judicial decisions, on the other hand, are a last resort. 50 Guidance at the international level must generally be exhausted before international criminal courts and tribunals will turn to decisions from national courts. 51

Scope and Methodology
From the discussion above, it is clear that international courts use a wide variety of legal sources upon which to base their decisions. Data was collected up to 2 May 2014. By this date, the ICC chambers had issued 119 decisions and orders 52 related to the Uganda matter and 455 decisions and orders related to the CAR matter. The Uganda and CAR matters were selected because they were two of the earliest -the first and third -situations referred to the ICC. The situation in the Democratic Republic of the Congo -the second matter to reach the ICC and the only other situation in which trials have commenced (and, in the case of Thomas Lubanga, resulted in a conviction) -was not chosen because the number of records generated (1,810) would have made a timely analysis unfeasible. Naturally, a higher number of records would result in superior data, and, in fact, a study that included all records from all situations would be preferable, but the author submits that the data presented herein -although limited -nonetheless provides meaningful insight into the ICC's use of precedent.
Put simply, the methodology used for this article to determine the ICC's referencing was to count the number of cases, treaties, articles, treatises and other sources that each chamber used as a basis for its decisions. The crucial element that made something 'precedent' was that it was used by an ICC chamber to support a decision. Put another way, each footnote of each order and decision was analysed to determine whether the sources cited therein were merely part of the procedural history of the matter (in which case, they were not counted) or whether they were being used by the chamber to assist in determining a point of law or in making a decision (in which case, they were). This determination -precedent or not -is a matter of judgment. At times, the distinction was a difficult one to make. The most challenging situations arose when references were made to sources that did not appear to be assisting the judges in their decision making but, at the same time, were not merely part of the procedural history of the case. In these situations, the referents have been designated 'mentioned precedent' rather than persuasive precedent. They were uncommon.
When an ICC chamber cited to a source more than once in the same order or decision, the source was only counted once. However, when both the majority and a dissenting opinion cited a source, it was counted twice. As a hypothetical illustration, if the majority opinion in the Pre-Trial Chamber's Decision on the Prosecutor's Application for a Warrant of Arrest cited XYZ v. HIJ case 10 times in its opinion, and a judge dissenting to the decision cited XYZ v. HIJ three times, this article would reflect two citations -one from the majority and one from the dissent.
When one ICC chamber mentioned that another ICC chamber used a sourcethat is, a chamber referred to another chamber's use of precedent -that source was only counted once (as the persuasive precedent of the first chamber that cited it). Additionally, when a chamber cited a source that in turn cited another source (for example, a Pre-Trial Chamber footnote stated: 'XYZ v. HIJ (citing to Smith v. Barney)'), only the first source (in this case, XYZ v. HIJ) was counted.
Precedent was collected from all types of opinions: majority, separate and dissenting. As mentioned above, like citations to the Rome Statute itself, citations to travaux preparatoires and other ICC-related materials (other than ICC judgments), such as the Report of the Preparatory Committee of the Establishment of an International Criminal Court and the United Nations Diplomatic Conference of Plenipotentiaries on the Establishment of an International Criminal Court, were not included in the data because they do not meaningfully reflect the Court's exercise of discretion to choose among sources. 53 The ICC chambers frequently cited the decisions of other international tribunals, such as the ICTR, the ICTY and the SCSL. Each judicial decision of these courts was counted, but different decisions in the same case were only counted once. In other words, if a judgment concerning the use of witnesses in the ICTY case of Prosecutor v. Dusko Tadić was cited, it would not be counted again if the warrant of arrest in Dusko Tadić was also cited in the same ICC record. Citations to statutes of these tribunals and courts were also included in the data, although it is worth noting that they were rare. When the ICC chambers issued identical decisions twice under different document identification numbers, the citations have only been counted once. 54 When a citation was made to an ICC record that was unavailable, it was excluded from the data unless the chamber issuing the record was ascertainable. Very few records were unavailable. 55

Data A Rulings and Precedents Generally
At the ICC, rulings are associated either with a situation generally or with a specific case. For instance, a ruling can be issued as part of the proceedings related to the situation in the CAR or in connection with a case against a particular defendant, such as Prosecutor v. Jean-Pierre Bemba Gombo. Figures 1A and 1B  Chamber. At the ICC, an Appeals Chamber will generally become involved in a matter before a Trial Chamber because issues from the Pre-Trial Chamber are appealed to the Appeals Chamber before a trial commences. Nonetheless, the order in this article reflects the more typical system in which appellate courts generally hear matters after (and are considered 'superior to') trial courts.
'references' refer only to references made to sources of law that were used as a basis for an ICC decision or point of law. 57 Figures 1C and 1D indicate the number of court records and citations in the situation in the CAR and the two cases that have been opened against CAR defendants. The ICC chambers issued significantly fewer records in the Uganda matter -119 compared to 455 in the CAR matter. The Situation in Uganda was brought before the ICC in March 2004, but the first defendant to be arrested, Dominic Ongwen, was not surrendered to ICC custody until 16 January 2015. As a result, the chambers have been less active. Though the Situation in the Central African Republic was referred to the ICC in December 2004 -approximately nine months after Uganda -Jean-Pierre Bemba Gombo, the former vice-president of the Democratic Republic of the Congo, was arrested in May 2008 for war crimes and crimes against humanity allegedly committed in the CAR and was transferred to the ICC in July 2008. This led to a sharp increase in court records. His trial is ongoing. The numbers of records and citations divided by chamber, shown below in Figures 2A and 2B, shed light on the different frequency with which each chamber references. Figures 2A and 2B include records with no references (sometimes these brief orders were only two or three pages long). This article specifies with each figure whether records with no references were included. In deciding whether to include or exclude records with no references, it has attempted to portray the data in the manner When records with no references are excluded, the number of average citations per record increases and the rankings change slightly, as indicated by comparing Figures 3A (including references) and 3B (excluding references). For instance, the Pre-Trial Chamber in the Uganda matter moves from lowest when including records with no citations to third highest when excluding these records. There are two factors to consider when comparing Figures 3A and 3B. First, different chambers have different percentages of records with no citations. For instance, in the CAR matter, only 10 per cent of the Pre-Trial Chamber records included references, in contrast to 31 per cent of the Appeals Chamber records and 49 per cent of the Trial Chamber records. Second, records with citations contain different amounts of citations. Thus, even though a Pre-Trial Chamber may have more records with no citations, those records it has with citations may have more per record than another chamber.
The Pre-Trial Chamber averaged slightly more than two citations per record less than the Appeals Chamber in the Uganda matter, while the Pre-Trial Chambers referencing. While pre-trial chambers also face a number of requests, objections and applications from the parties, they would presumably require fewer references than the matters raised at the appeals level. The data, however, indicate that, at least in the CAR matter, the resolution of issues raised before the pre-trial judges may be requiring as much citation to legal authorities as those at the appellate level. In the CAR matter, the Trial Chamber averaged fewer precedents per record than the other two chambers. As indicated by Figure 3B, the combined chambers presiding over the Uganda matter have averaged slightly more citations per record (excluding records with no references) than the combined chambers presiding over the CAR matter. 58

B Use of Precedents over Time
The ICC's use of precedent over time generally reflects the progress of the cases before it. As a case develops from referral, to arrest warrant, to arrest, to trial, the number of court records issued rises. This correlation makes sense because as issues before the court become more complex and the case more active, the chambers must look to applicable legal sources to resolve them. Figures 4A, 4B, 4C, 4D and 4E illustrate the use of references over time in the Uganda and CAR matters (key events are noted in the bubble boxes). 59 58 Figure 3B  Naturally, the rise and fall of referencing reflects only somewhat the number of records issued because some records contain many more references than others, even when the data is restricted to records that contain citations. Somewhat predictably, the number of references used at the pre-trial level far exceed those at the appeals level shortly after the case commenced because the parties had not yet appealed any pre-trial decisions, while citations at the appeals level then rose sharply once the pre-trial decisions had time to be raised on appeal. Figures 4A and 4B end at 2012 because no records were issued between that year and the date on which data collection was completed.
The data for the chambers handling the CAR matter reflects similar trends. 60 The number of records at the pre-trial and appeals levels rose slowly and steadily after the investigation 60 Records without citations were excluded.

C Nature of Sources
Perhaps the most useful information from the collected data, at least for the advocates appearing before the ICC, concerns the nature -that is, the types and the identities of the specific sources of law -of the citations used by the ICC. Recall that Article 21 requires the ICC to apply, first, the Rome Statute; second, the Elements of Crimes and the Rules of Evidence and Procedure; third, treaties and international law; and, fourth, general principles of law derived from national legal systems. The Court may, but need not, apply principles and rules of law as interpreted in its previous decisions.
Which sources of law do the ICC chambers prefer? Do the sources reflect the hierarchy established by Article 21 of the Rome Statute? When looking outside the ICC's own case law, do ICC judges prefer certain international -or even national -courts and tribunals? Figures 5A and 5B illustrate the types of sources of law referenced by the chambers hearing the Uganda matter, indicating the chambers' different reliance on case law, treatises and journal articles, treaties, analogous rules 62 and so on.
Case law is clearly the dominant source of law for the ICC chambers in the Uganda matter. 63 Over three quarters of the references made by both the Pre-Trial and Appeals Chambers are persuasive, mentioned and distinguished case law. 64 Treaties, the second most important category under Article 21 of the Rome Statute -after only the ICC's own statute and rules -comprised only 2-3 per cent of the citations. Treatises and journal articles, a source not expressly authorized by Article 21, although likely falling under the umbrella of Article 21(1)(c) because they provide applicable general principles of law, comprise several times as many citations as treaties. 65 62 Analogous rules in the Uganda records were the Rules of Procedure and Evidence of the ICTY and ICTR. 63 The case law in Figures 5A and 5B has not been separated based on the particular court or whether the decision is external -that is, other than an ICC case -or internal. Figures 6A and 6B show these distinctions. 64 While distinguished cases are not properly a source of law, they are significant in that they are a legal record that the ICC judge believed was significant enough to have to distinguish. Accordingly, they have been included in the data. 65 ICJ Statute, supra note 27, Art. 38(1)(d) authorizes the Court to apply 'the teachings of the most highly qualified publicists of the various nations, as subsidiary means for the determination of rules of law'. The Rome Statute does not have an analogous provision. As indicated by Figures 5C, 5D and 5E, the use of case law as precedent is even more pronounced in the CAR records, ranging from 72-92 per cent (excluding cases distinguished) of all citations. Treaties and conventions, such as the International Covenant on Civil and Political Rights and the European Convention for the Protection of Human Rights and Fundamental Freedoms, continue to be cited, albeit in relatively insignificant numbers. 66 While Figures 5A-5E divide sources of law by type, Figures 6A and 6B divide case law precedents by court. They demonstrate the predominance of ICC decisions in the case law cited. There is apparently little reluctance to look inward, even though the ICC is still a relatively new court with only two successful convictions. 67 There is, of course, nothing wrong with citing previous ICC decisions, as per Article 21(2). In fact, citing prior decisions leads to uniformity and predictability in the law. Nonetheless, the citation rate of ICC decisions is in marked contrast to the citing of the mandatory sources of law listed in Article 21(1)(b) -applicable treaties and the principles and rules of international law -and Article 21(1)(c) -general principles of law derived by the Court from national laws of legal systems of the world, including, as appropriate, the national laws of states that would normally exercise jurisdiction over the crime (which would be either Ugandan or CAR laws). This does not mean that the ICC is referencing incorrectly, but it may indicate that the higher categories of sources of law in Article 21 are less usefulor, perhaps, less frequently useful -than lower sources. In both matters, decisions of the ECtHR, the ICTY, the ICTR, the SCSL and the Inter-American Court of Human Rights also featured prominently. Decisions of the European Commission of Human Rights, while not technically a court, were also cited in both the Uganda and CAR matters.  At the outset, one of the goals of this research was to determine whether the ICC, when referencing national judgments, preferred precedent from Western (European and US) legal systems. In other words, in a prosecution of a Ugandan defendant, for example, does the ICC prefer to reference Western case law or Ugandan case law? The issue of ICC bias has become increasingly important as all of the Court's official investigations have been opened in African countries, 72 and African leaders have accused the ICC of targeting Africans while exempting Western leaders from prosecution. 73 Article 21(1)(c) of the Rome Statute arguably encourages the Court to reference general principles of law derived from the laws of the defendant's country of origin. It provides that, if those sources enumerated in Articles 21(1)(a) and 21(1)(b) are not applicable, the Court shall apply: general principles of law derived by the Court from national laws of legal systems of the world including, as appropriate, the national laws of States that would normally exercise jurisdiction over the crime, provided that those principles are not inconsistent with this Statute and with international law and internationally recognized norms and standards. 74 The drafters of the Rome Statute could have omitted the language 'including, as appropriate, the national laws of States that would normally exercise jurisdiction over the crime' because, using the example again, Ugandan laws would clearly be included in the immediately preceding phrase 'national laws of legal systems of the world'. The drafters chose, however, to include this language, indicating perhaps not only that the Court should not ignore Uganda's laws but also that it should look to general principles of law derived from Uganda's laws before those of other states. To do so would be sensible since a Ugandan defendant would justifiably expect principles from Ugandan laws, when appropriate and not inconsistent with the Rome Statute and international laws, norms and standards, to be applied before principles derived from other countries' laws.
The collected data reveal that, at least in the Uganda and CAR matters, the ICC perhaps just slightly prefers Western laws. The ICC Chambers presiding over the Uganda matter referenced one English, one US, one South African, one Ugandan and two German laws. The Chambers adjudicating the CAR matter referenced one German and three English laws. The significance of these figures should not be overstated. First, the overall amount of domestic decisions referenced (six and four, respectively) is too small to be statistically meaningful. Second, it is not possible to ascertain from the references whether the Chambers considered Ugandan and CAR laws (or other African laws) but determined that they were either inapplicable or inconsistent with international laws, norms and standards -in which case, their failure to reference these laws would not indicate bias.

D Self-Citations
A practice of some interest is not merely the citing of the decisions of other ICC chambers but also the citing of one's very own decisions. Citing one's own decisions is common and perfectly acceptable in domestic legal systems 75 and presumably in international courts and tribunals as well. Judges of a relatively new court, such as the ICC, would be doubly justified in citing their own decisions. After all, they have little ability to cite other judges on the same court simply because there have not been many other judges to cite. Nevertheless, citing to different ICC judges' opinions should be viewed as healthy because it encourages diversity of precedent and, at least, gives the appearance that judges are not merely using their own prior opinions to buttress their current opinions. Figures 7A, 7B, 7C  Judge Ekaterina Trendafilova of Bulgaria led self-citing at the Pre-Trial Chambers for the CAR matter, averaging a little more than one self-citation in three records. The statistics for Judges Saiga, Slade and Steiner are insignificant because they issued so few records in the CAR matter at the Pre-Trial level -2, 1 and 2 respectively. 77 The self-citing average increased at the Trial Chamber, where it ranged from slightly over one-half citation to slightly over one-and-a-half citations per record, and at the Appeals Chamber, where it ranged from nearly one-half citation to one-and-a-third citations per record. 78 The data reflects average self-citations per record, including records without citations. In Figure 7A, the numbers of records issued, in the order of the judges listed, were 88, 69, 63, 2, 16, 32, 1 and 2 respectively. 78 The data reflect average self-citations per record, including records without citations. In Figure 7B, the numbers of records issued, in the order of the judges listed, were 109, 150, 109, 41 and 41 respectively and in Figure 7C, 68, 71, 68, 70, 62, 3, 3 and 9 respectively. As with all statistics in this study, self-citing averages may reflect a number of factors that this article does not attempt to address. For instance, certain judges may have participated in seminal cases, making them more likely to later cite these landmark opinions. Judges who have higher averages may have been confronted with more legal issues that they had already addressed in earlier decisions than other judges. 79 As indicated in Figure 7D, overall self-citing in the CAR case -the only case in which self-citing was analysed -was highest in the Appeals Chamber, where it was over three times the average in the Trial Chamber and over ten times the average in the Pre-Trial Chambers.
The data in Figure 7D, as opposed to that in Figures 7A, 7B and 7C, show the average self-citations per chamber, excluding records without citations. As discussed above, including or excluding records without citations significantly impacts data. In Figures 7A, 7B and 7C, records without citations were included to show how often judges were self-citing among all their records. For instance, if a judge issued ten records, nine of which had no citations and one that had twenty-five citations, one of them being a self-citation, the data shows the judge as having a 0.10 self-citation rate (one citation in ten records). In contrast, the rate of a different judge who only issues one record that has twenty-five citations, one of them being a self-citation, would be 1.00. So long as a judge is only compared to other judges within the same chamber level, this method reflects their self-citation rate fairly, considering all records issued. Excluding records without citations would, in this author's view, lead to deceptively high self-citation rates. On the other hand, Figure 7D compares self-citing between chambers. Since Pre-Trial Chambers issue many more records without citations than the other two chambers, these records have been excluded from Figure 7D so as to avoid a deceptively low self-citation rate for the judges of the Pre-Trial Chambers.

Conclusion
Referencing at international courts remains a largely neglected, but crucial, aspect of emerging international jurisprudence. The patterns revealed by the data collected and analysed in this study contribute to a better understanding of how, at its core, the ICC arrives at its judgments. Each time one of the Court's chambers references -or declines to reference -a source of law, it is making a decision that, in the aggregate, has real consequences for the parties before it. The data in this article may prove most useful to the Court itself. While judges are of course well aware of the precedent they use on a case-by-case scale, they may be less cognizant of wider referencing trends. The referencing patterns reveal perhaps latent, though not necessarily prejudicial or improper, preferences that are not obvious during the day-to-day process of judging.
The ICC is a particularly worthwhile subject of study because it is a relatively new court -its founding statute was adopted in 1998 and entered into force in 2002 -which may prove more amenable than more established tribunals to evaluating the strength and credibility of its decision-making procedures as well as in its willingness to make necessary adjustments. Parties appearing before the court -prosecutors, defence counsel, legal representatives of victims and amicus curiae -may also find the data useful. These parties may be able to increase the effectiveness of their legal submissions by referencing those source materials that are most frequently utilized in ICC judgments and, when appropriate, can even tailor their use of precedent to the preferences of particular chambers. 80 This article is only a beginning. Additional data must be collected and analysed to provide a complete picture of referencing patterns at the ICC. The Court has commenced investigations (in addition to those of Uganda and the CAR) into situations in the DRC, Mali, Libya, the Republic of Côte d'Ivoire, the Republic of Kenya, Darfur and Sudan. Different judges hearing different cases and confronted with different issues may well turn to different sources of law. As the Court matures, referencing will likely shift to an even greater extent to internal ICC case law. 81 With increased decisions and a greater diversity of opinions to cull from, there will be less of a need to look elsewhere for precedent. Likewise, self-citing will almost surely diminish.
The data from this study can lead to further areas of exploration. For example, more analysis on the Court's use of other sources of law, such as journal articles and texts, or additional evaluation of the Court's citation to other international tribunals, could clarify the importance of these sources. A comparative analysis with the referencing rates of other international courts would assist in determining whether the ICC's referencing is unusual or conventional. Perhaps one of the most interesting areas of future study will be to see how the Court's referencing changes over time since the Court, in the context of the overall development of international law, remains a relative newcomer. This study and other similar data-driven analyses of the ICC may also counterbalance some of the recent criticism of the ICC by assessing the Court's work from an objective, non-political perspective. 80 Although the data in this article does not address referencing by specific judges other than in connection with self-referencing, the identities of judges from a particular chamber can be discerned without great difficulty. 81 See Borda, supra note 1, at 14-18; Borda, supra note 37, at 296.