Trouble in Paradise: Expanding Applications of the Getty Thesaurus of Geographic Names ® to Enhance Intellectual Discoverability of Circum-Caribbean Materials

This article examines how the Circum-Caribbean region ’ s cultural and geographic complexities make it difficult to describe or index relevant archival materials using the mainstream authority controls used in galleries, libraries, archives, and museums (GLAMs). This difficulty stems from the fact that authority controls utilised by GLAMs are primarily created by North American or European authorities and, therefore, have Western-centric views imbued with colonialist overtones. When these systems are used to catalogue, index, or describe Circum-Caribbean-related collection materials, a tension arises: a system with a white, Euro-American perspective is applied to material reflective of a significantly multicultural place, culture, subject, and population. The rigidity of controlled vocabularies and their applications — which typically follow specific indexing methodologies — cannot accommodate the fluidity necessary to accurately denote the complex Circum-Caribbean region, especially with regard to geographic indexing. This article demonstrates the difficulties that emerge from trying to delimit and define the Caribbean region; provides an abbreviated analysis of the Circum-Caribbean ’ s representation in the Getty Thesaurus of Geographic Names ® (TGN), which mirrors the difficulties of defining and delimiting the region; and presents a case study in which the West Indian Postcard Collection at Cambridge University Library was indexed using augmented applications of the TGN. The research presented in this paper supports the theory that employing both general and specific indexing strategies creates enhanced access to Caribbean-related collection materials by enabling regional, sub-regional, and territorial/national avenues to retrieve collection materials.


Introduction
Galleries, libraries, archives, and museums (GLAMs) have historically suppressed, overlooked, and excluded non-Western-centric perspectives in collecting, describing, and indexing collection materials. Many institutions and scholars are increasingly critiquing collections and cataloguing practices, striving to find ways to decolonise 1 and reconcile them in an effort to visibilise narratives that have been historically suppressed in GLAMs (Buckley 2008;Campt 2012Campt , 2017Turner 2015Turner , 2020Bosum and Dunne 2017;White 2018). However, very little academic work addresses the importance of decolonising and reconciling with 1 Following my presentation at the Museums Assocation of the Caribbean's 2019 conference, the discussion among Caribbean peers in the field made it clear that there is growing disagreement about what the term decolonisation means and if it is appropriate or inflicts further harm. When I use this term or its derivatives, it is from the distinct perspective that decolonial work is not about "undoing" or "erasing" the colonial past. Rather, I approach decolonial work as "the effort to interrogate and transform the institutional, structural and epistemological legacies of colonialism" and "an additive process which heightens the accuracy and completeness of our knowledge of the world" (Meghji 2021). collection materials pertaining to the Circum-Caribbean despite the cataclysmic colonisation that dominated the region for centuries. 2 It is in this context that this paper examines the inefficiency of rigid mainstream authority controls used in GLAMs, such as controlled vocabularies and specific indexing methodologies, and demonstrates how this rigidity stunts intellectual discoverability of Caribbean-related collection materials. This paper will demonstrate that North American-and European-created authority controls used in GLAMs need to be investigated to understand how their creators' biases suppress non-Western-centric cultures during the cataloguing of collection materials and suggests a dual-indexing approach to improve discoverability of Caribbean-related materials.
The first half of the article provides a brief history of modern Anglo-American information organisation and standardisation in GLAMs, introduces the traditionally binary concepts of general and specific indexing, and demonstrates the complexity of defining and delimiting the Circum-Caribbean region. The second half of the article presents an abbreviated analysis of how the Circum-Caribbean is depicted within the Getty Thesaurus of Geographic Names® (TGN) and discusses a case study conducted by the author in which the TGN was applied to a postcard collection held at Cambridge University Library to determine how different methods of application impacted intellectual discoverability. As will be further explained in this article, the research presented here measures intellectual discoverability as the number of geographic place names that are applied as indexed terms and, consequently, act as controlled access points. 3 Ultimately, this paper aims to highlight the less-than-accurate ways in which Circum-Caribbean materials are often organised in GLAMs' collection catalogues. It proposes a methodology that can be used to critique authority controls to make them more adaptable and suitable to culturally specific materials. Though this paper is focused on doing so through a Circum-Caribbean lens, the same methodology could be applied to these standardised authority controls from any inaccurately or underrepresented place or culture.

Information Organisation and Standardisation
Before analysing existing information organisation systems, it is important to understand the earliest predecessors of such systems, the objects for which they were created, and subsequent developments. For almost two centuries, efforts to systematise information organisation and archival description have been driven largely by librarians and archivists focused primarily on developing authority control systems for bibliographic standardisation of textual materials (Enser 1993). Sir Antonio Panizzi's plan to organise books in the British Library with a new catalogue from 1839 into the early 1840s is lauded as the start of the modern Anglo-American history of systematic information organisation in the tradition of descriptive and subject cataloguing (Svenonius 2000;New World Encyclopedia contributors 2016). 4 His argument for a catalogue to group "like" items and differentiate among similar ones indirectly referenced bibliographic objectives (later explicitly stated by Charles Ammi Cutter), and his ensuing efforts laid the groundwork for the birth and development of all major subsequent cataloguing systems used worldwide today (Svenonius 2000;New World Encyclopaedia contributors 2016). However, the "book catalogue" format of Panizzi's era was onerous: it consisted of one or more large volumes in which bibliographic information about works (initially books or literary items) was entered by hand. It required empty spaces to be left where future bibliographic listings could be inserted. Significantly, Panizzi recognised the importance of extensive "See" and "See also" cross-references to demonstrate relationships between certain items in order to assist in navigation and collocation (Svenonius 2000).
However, it was not until the second half of the nineteenth century that American librarian Charles Ammi Cutter revolutionised the field of bibliographic information organisation and introduced the card catalogue (Svenonius 2000). Unlike book catalogues, bibliographic information of a work was detailed on a card-usually one card per work. This was an improvement on the book catalogue as it allowed the card catalogue to grow as needed without the necessity of leaving blank spaces for newly acquired items. The card catalogue system also has a certain degree of adaptability; it is much easier to remove cards for deaccessions, and the catalogue can be reorganised if the collection is broken up into smaller departments later on, whereas a book catalogue would have to be cut into pieces to do this (and, even then, items may be listed on both sides of a page). In 1901, the Library of Congress began distributing card-catalogue copy to American libraries, a milestone towards economising bibliographic effort and "recognising the possibility of universal bibliographic control through standardization" (Svenonius 2000). This can be seen as an early instance of cooperative cataloguing. Cutter's system also included an author index and a "classed catalog"-in short, a precursor to a subject index (Svenonius 2000;New World Encyclopaedia contributors 2016).
Card catalogues eventually transitioned from analogue to digital to online. Geoffrey Bowker and Susan Leigh Star (1999, 2) state that these control systems "were jolted in the twentieth century by information explosions, the computer revolution, the proliferation of new media, and the drive toward universal bibliographic control." Each catalogue type has its own set of unique challenges to achieving universal bibliographic control. According to Elaine Svenonius, the relational and syndetic structures that were clear in book and card catalogues were eroded in the eventual transition to online catalogues, causing a "steady deterioration in the integrity of bibliographic structures" and "an undermining of bibliographic objectives," resulting in less user-friendly catalogues (2000,(63)(64).
More recently, GLAMs and archival associations have contributed to the development, implementation, and maintenance of so-called best practices for globally standardised archival description methods. Such contributions include the Anglo-American Cataloguing Rules (AACR, first developed in 1967), the Bureau of Canadian Archivists' Rules for Archival Description (first published in 1990), the General International Standard for Archival Description (ISAD[G], first published in 1994), and the International Federation of Library Associations and Institutions' (IFLA's) Fundamental Requirements for Bibliographic Records (FRBR, first published in 1998). The archival description systems that are used in cataloguing today are often complemented by more specific authority controls-for example, geographic thesauri, name authorities, or subject heading indexes such as the Library of Congress Subject Headings (established in 1898), the Canadian Subject Headings (begun in 1968; first published in 1978), and the Getty Thesaurus of Geographic Names® (begun in 1987; first published in 1997).
At the core of these complementary systems are a controlled vocabulary and syntaxes that govern the many authority records within the authority control and their relationships to each other, the creation of new records, and the implementation of the control system. But their controlled vocabularies are, by nature, often exclusionary. Bowker and Star (1999) emphasise how the construction of each standard or categorisation system valorises one perspective above others, and they urge us to consider how such choices are made. Controlled vocabularies seldom meet the needs of all institutions or collections or permit the necessary flexibility to accurately describe multifaceted cultures or subjects. As a result, they often undermine intellectual discoverability and efficient collation of items. Furthermore, disregarding the needs of a collection or institution in favour of industry-prescribed standards has the potentially unintended effect of further colonising collections and oppressing certain viewpoints if the standards applied do not consider the nuances of the materials being catalogued.
Cataloguing librarian Linnea Marshall (2003) explores the classic arguments of specific versus general cataloguing by recounting debates amongst librarians during the latter half of the twentieth century, carefully considering the relativity of specificity as it relates to indexing vocabularies, recall and precision, and the value of both general and specific headings in subject searching. Marshall concludes that the principle of specificity does not serve all catalogue users since it "favors not only the narrowly-focused researcher over the comprehensive researcher [who wants recall], but also it favors the narrowly-focused and knowledgeable specialist over the narrowly-focused but less knowledgeable researcher" (2003,69). The author discusses several projects that support the idea that "assigning both specific and generic subject headings to a work would enhance the subject accessibility for the diverse approaches and research needs of different catalog users" (Marshall 2003, abstract). If regional place names are considered in place of "general headings" and territorial/national place names in place of "specific headings," employing a dual-indexing strategy by applying both general (regional) and specific (territorial/ national) place names could simultaneously serve a user seeking all materials from a region and another user searching for a single territory. This article argues for this dual-indexing approach, noting, however, that the primary downfall to this strategy is that it could become more difficult for users to recall materials that deal comprehensively with the region in general, excluding territory/nation-specific materials.

Defining and Delimiting the Circum-Caribbean
One of the major issues in describing and indexing Circum-Caribbean materials in GLAMs is the problematic and inconsistent use of nomenclature relating to the Circum-Caribbean in authority controls-for example, the unexplained and inconsistent use of Caribbean and West Indies in the collection catalogues of multiple institutions. A considerable amount of Caribbean scholarship engages with the difficulty of what to name Caribbean places and how to define or delimit the Caribbean area, but institutions (especially those outside the region) take these place names and their delimitations for granted. Eurocentric narratives of the Caribbean area are well known; however, works by Caribbean scholars and authors provide historical narratives that are more nuanced and better reflect the region's geographical, socio-political, cultural, and historical complexities. This section draws on primarily Caribbean authors to discuss how cataloguing Caribbean-related materials is complicated by the ambiguous origins of place names, the notion of Creolisation, and multiculturalism and multilingualism in the region. These factors all influence attempts to delimit the Caribbean area as a region and space and challenge attempts to standardise Caribbean place names. This context is crucial for indexing materials related to the region and underpins my argument for a dual-indexing approach to applying place names.
Sources on the history of the Caribbean offer different accounts of the origins of regional place names. For example, John Parry, Philip Sherlock, and Anthony Maingot (1987, 6) explain that when Spanish explorer Christopher Columbus reached the islands, he mistakenly claimed he had reached the East Indies, soon dubbing the islands the West Indies once the Spaniards realised their mistake. Meanwhile, ethnic studies scholar Tony Castanha (2011, xvi) argues that the term Caribbean is derived from the name of one of the region's Indigenous populations, the Caribs. In this sense, these place names could be said to hold different-colonialist versus self-representative-connotations. Antonio Gaztambide-Géigel (2004) summarises multiple competing hypotheses about how the term Caribbean originated to demonstrate the ambiguity of the name's true origins. He states that the word Caribbean was not widely used until the United States imposed it but now it is generally accepted as the name of the region (Gaztambide-Géigel 2004, 128), a direct contradiction to Castanha and others' hypothesis that it was self-derived by the Caribs. Gaztambide-Géigel argues that the term Caribs was used by Christopher Columbus to refer to one of the region's Indigenous populations (2004,129). The indeterminate origins of these terms pose problems for cataloguing in GLAMs which already use Caribbean and West Indies inconsistently. While critics have proposed that Caribbean should be the term used, can the term Caribbean be considered more appropriate than West Indies in an anti-colonial framework if it was also imposed by Euro-American outsider populations? Furthermore, according to Gaztambide-Géigel, geographic terms such as Lesser Antilles, West Indies, and Leeward and Windward Islands are also imprecise because although different colonising powers may have used the same terms, they were not always referring to the same islands (2004,. The uncertain origins of all these place names make it challenging to establish standardised terms for cataloguing purposes and make it difficult to determine which names (if any) are less colonial and perhaps, therefore, more reflective of the region's self-derived identity.
Cataloguing Caribbean-related materials is further complicated by the concept of Creolisation, a term introduced by Edward Kamau Brathwaite in 1974 to describe when elements of different cultures are blended together to create a new culture. Creolisation is explored in depth by the canonical Martinican poet-novelist-theorist Édouard Glissant (1969Glissant ( [1989), who coined the term Antillanité, or "Caribbeanness." Both concepts remain pertinent to Caribbean studies and, I argue, should be considered and used as a framework to influence how Caribbean-related collection materials are catalogued. Using a Creoleinformed approach would provide space to necessarily nuance the Caribbean place names used, but would not easily conform to the rigid metadata standards implemented across GLAMs that strive to enforce strict and simplistic categorisation, with little to no flexibility to account for more nuanced or multifaceted concepts, such as Creolisation or multilingualism. As Gaztambide-Géigel (2004) notes, one of the most prominent challenges facing Caribbean GLAM professionals within the region is how to make Caribbean-related materials accessible for a multilingual regional audience. Due to its multinational colonial past, English, Spanish, French, and Dutch are all official languages in different parts of the region. The existence of numerous official and unofficial Creolised languages like Haitian Creole, Papiamento, and patois only further complicates attempts to standardise Caribbean place names (Sanabria 2007). 5 In addition to struggling to name Caribbean places, Caribbean studies scholars also struggle to define or delimit where, exactly, constitutes the Caribbean area. Ileana Sanz discusses the "oscillating pendulum" between regionalism and nationalism within the "Circum-Caribbean" or "Wider Caribbean," which she defines as "a space which includes the insular Caribbean, together with the northern coastal states of South America, Central America, and the Caribbean coast of Mexico" (2009, 1). Despite noting that relevant literature remains sparse because Circum-Caribbean theory is still fairly new, Sanz demonstrates how challenging it is to extricate the Caribbean from other historically linked areas. For example, distinguishing Latin America from the Caribbean is difficult due to their similar geographical and historical colonial contexts, used by "visionary and . . . founding father of Latin America" Simón Bolívar to invent "his [Latin] America" (Sanz 2009, 4-5 ), which included island territories such as Puerto Rico, Cuba, and the Dominican Republicplaces that are, geophysically, part of the Caribbean archipelago. Likewise, distinguishing the Caribbean from Marcus Garvey's "Black America" is difficult since Garvey sought to integrate all people of African origin-be they in the Caribbean, the United States, or Africa (Sanz 2009, 10). Sanz's article is situated within the discourse of space and place, topics also addressed by Latin and Caribbean studies professor Michelle Stephens (2013) and Caribbean and Africana studies professor Carole Boyce Davies (2013). Davies suggests that "Caribbean Spaces" are ever expanding and without boundaries, a phenomenon signified by the Caribbean diaspora, "re-creations of Caribbean communities following migration" (2013, 1), while Stephens questions whether the Caribbean transcends geographical place (2013,8). The tensions between using insular versus archipelagic and nationalist versus regionalist approaches are clearly visible in both authors' works.
Overall, complex historical integrationist approaches to identity and sovereignty still exist and impact how the Caribbean is represented in institutional collection catalogues. The origins of terms like Caribbean and Windward Islands remain unclear, and the multiple languages spoken in the region further complicate attempts to catalogue Caribbean-related materials in a way that will serve their communities. Below, I will demonstrate how these complexities make it difficult to conform to Euro-American-created authority controls used by GLAMs in their collection cataloguing. Using an applied case study that focuses on the Getty Thesaurus of Geographic Names® in conjunction with a postcard collection at Cambridge University Library, I outline how a dual-indexing approach can alleviate some of these cataloguing challenges.

Analysing the Getty Thesaurus of Geographic Names®
I chose the Getty Thesaurus of Geographic Names® (henceforth referred to as the TGN) for analysis since it is the primary geographic authority control used in the Royal Commonwealth Society (RCS) department at Cambridge University Library, where my case study collection resides (Paul 2017, Section 6). The TGN is a "structured [resource] that can be used to improve access to information" (Getty Research Institute 2019) and contains over 2.5 million place names. Like all other Getty Vocabularies, it is constructed to enable its use in linked open data (LOD) and released as LOD to promote making "knowledge resources freely available to all." The TGN's development claims to be "focused on the historical world and places necessary for cataloguing and discovery of visual works" and cites "adding archaeological sites, lost sites, and other historical sites, particularly Pre-Columbian places and places in Asia, Middle East, Africa, and others" 6 as a core focus for its development (Getty Research Institute 2019). Primary users of the TGN include art museums, special collections, art libraries, archives, visual resource collection catalogers, bibliographic projects concerned with art, researchers in art and art history, and the information specialists dealing with the needs of these users. The TGN is available to users in a variety of formats. Users can search the thesaurus online (the primary method used for this research) and download datasets in various formats from the Getty Research Institute's website.
In order to analyse the TGN's representation of the Circum-Caribbean, I visually charted all authority records pertaining to the Circum-Caribbean in the TGN to a national or territorial level using software called Creately. This allowed me to review all relevant authority records simultaneously and made it easier to analyse the thesaurus's successes and failures by seeing the relationships between authority records all at once. In this way, the authority control's representation of the region could be viewed as an incomplete puzzle, but one that is complete enough to begin envisioning what the missing pieces might look like, where they might fit, or which pieces have been misplaced. I then analysed this visual chart in an effort to determine how well the TGN represents the Circum-Caribbean, asking questions like: what names are included or excluded? What sub-regional groupings are represented? Who are the main sources providing information about this region? This methodology provides a framework to approach and analyse other authority controls from a Circum-Caribbean perspective and can be adapted to investigate the representation of any other place, culture, or concept in a chosen authority control.

TGN at Cambridge University Library
Before analysing the TGN, it is important to explain how it is integrated and used at Cambridge University Library (the UL), as this context significantly impacted my analysis. Many institutions using the TGN as an authority control in their cataloguing workflows have TGN data incorporated within their collection management system, obtained from third-party vendors. Indexing using the principle of specificity, cataloguers identify and select the most specific and relevant location of an object from the list of TGN data in the collection management system's respective field, and the hierarchical relationship(s) are automatically generated from the LOD. However, not all institutions can support this embedded functionality. At the time of this research, the UL did not have the capacity to incorporate TGN datasets into its collection management system. Instead, cataloguers manually identified and assigned index terms by referring to the library's identified authority control lists-of which the TGN is one. This manual application process is not ideal because it requires substantially more time for the cataloguer and creates ample room for human error. This research takes into consideration how this manual application at the UL can affect users.
This research was carried out in early to mid-2020, prior to the UL's transition from a digital repository system called JANUS to the current platform, ArchiveSearch. The user interface represented in screenshots within this paper has thus changed, and some of the issues highlighted here may have been fixed during this system migration.

Physical and Political Approaches to Geography in the TGN
The TGN attempts to label places beyond strictly geophysical boundaries. The standardised place names, called preferred names, are classed by place type as either physical features or political entities. This categorisation of place type can be seen in all levels of the TGN: from directly below World (facet) at the upper level of the hierarchy, all the way down to localities within a specific territory or nation. If a user sees the text [view physical features], it means they are currently viewing political entities in a hierarchy, and vice versa. These two main branches can be broken down into further place types that are indicated as qualifiers in brackets following the place name-for example: (island), (island group), (nation), and (dependent state) (see Figure 1).
The TGN's acknowledgment that there are different-for example, physical and political-approaches to defining or classing places is commendable. However, in so doing, some places end up with duplicate or near-duplicate index terms to reflect these different approaches, but with a near-identical or seemingly identical name. For example, a physical feature name like Jamaica (island) and a political entity name like Jamaica (nation) both refer to the island country Jamaica in the Caribbean. In the back end of a cataloguing system, the cataloguer might index an object using both Jamaica (island) and Jamaica (nation). However, the place type-(island) or (nation)-might not be visible on the front-end user interface of the collection management system, depending on the software being used. This was the case with the RCS's online catalogue, and it is problematic for two reasons. First, it can easily confuse the user since (without a visible place type qualifier) it appears as though there are multiple identical headings for the same place in the index list-e.g., Jamaica and Jamaica. This can be difficult to navigate as a user. Further, if consistent cataloguing methods are not performed, seemingly duplicate or near-identical index terms can fracture the index's ability to perform accurate collocation-for example, if some objects are catalogued with Jamaica (island), others with Jamaica (nation), and others with both index terms.
Similar issues arise when one place is indexed with near-identical index terms-for example, Bahama Islands (island group) and Bahamas (nation) (Figure 2). A general user, who is not an expert in cataloguing or indexing, will likely be confused as to why there are multiple terms that seem to denote the same place and would need to conduct a verification process on any retrieved results to ensure that they have collocated all relevant materials and/or not obtained duplicate results for their query. Therefore, while considering both physical and political concepts of geography can expand the breadth of potential research avenues of a collection's holdings-at least for the Circum-Caribbean region-it is clear that further thought and consideration is necessary to determine how near-duplicates within an authority control can be reduced to enable more consistent implementation by the cataloguer and better functionality for the user.

Polyhierarchical Relationships and Multiple Index Paths
The polyhierarchical nature of the TGN leads to duplicated names in another way, whereby the same place can be traced through several different index paths (see Figure 3). Each TGN authority record provides the term's hierarchical position (what I refer to as the "preferred hierarchy") as well as additional parents ("additional hierarchies"). However, there is no apparent explanation as to how the Getty determines the default (preferred) hierarchy. According to the Getty's application guidelines, the most specific place should be selected from the preferred hierarchy path when indexing materials. However, this begs the question: why bother having the other hierarchies if they are not recommended for use? It seems as though each institution or department using the TGN would be best served by creating a detailed cataloguing handbook to guide all staff (and/or volunteers) in applying the TGN in a consistent manner-including if staff should only index according to the Getty's guidelines or should expand their indexing strategies to include all hierarchies and all unique terms from all hierarchies combined. Since the thesis of this Gooding: Trouble in Paradise paper is that a dual-indexing strategy incorporating general and specific (regional and territorial/national) subject heading approaches would serve more users, including all terms from all hierarchies is beneficial because it increases the number of controlled access points that could allow for regional, sub-regional, or territorial/national levels of result retrieval.
If a cataloguer consulted only one path (the preferred hierarchy) for a place that is referenced by multiple possible paths (additional hierarchies), they will potentially omit other index terms-unique to the additional hierarchies-that could enable additional discoverability and collocation for some users' queries at regional or sub-regional levels. For example, there are three index paths that could be used to index the Caribbean territory Anguilla. These are:

World (facet) > British West Indies (general region) > Anguilla (dependent state) 7 2. World (facet) > West Indies (archipelago) > British West Indies (general region) > Anguilla (dependent state) 3. World (facet) > North and Central America (continent) > Anguilla (dependent state)
If a cataloguer used only the third index path, they would omit the additional terms British West Indies (general region) and West Indies (archipelago) found in paths one and two, terms that could help a user collocate items at a broader regional or sub-regional level. Worse still, if a cataloguer simply (manually) indexed using the term Anguilla (dependent state), the omission of any other terms would only enable a researcher to gather information on a national level, excluding Anguilla-related materials in any regional or sub-regional searches. It seems unlikely that all multiple index paths can be eradicated, but there are some that seem unnecessary, such as the example illustrated below for the island nation Dominica (see Figure 3).
It is unclear why both hierarchies need to exist for Dominica (island) and why Windward Islands (island group), in the additional parents hierarchy, is not simply incorporated into the hierarchical position hierarchy since both that term and Dominica (island) fall below North and Central America (continent). Thus, a proposed single path including all terms is:

World (facet) > North and Central America (continent) > Windward Islands (island group) > Dominica (nation) > Dominica (island)
Such a solution may solve some instances of multiple index paths, but not all. For example, Figure 3 shows only two of at least six possible index paths leading to Dominica. Furthermore, a simple but frustrating issue with the TGN's hierarchies is the sheer number of inconsistencies that can be found within the structure itself. For example, it is unclear why the Greater Antilles (island group) is hierarchically below the continental level North and Central America (continent) but that the same is not applied to Lesser Antilles (island group) since they are both listed under West Indies (archipelago) in an additional parent-and West Indies (archipelago) is part of North and Central America (continent) (Figure 4).
Another structural inconsistency is seen for the nation Antigua and Barbuda: it is unclear why Antigua is listed as a physical feature and Barbuda is listed only as a political entity when, in reality, it is also its own distinct island ( Figure 5). Should Barbuda not also be represented as a physical feature, namely Barbuda (island)?
Another structural issue is the Getty's attempt to hierarchically represent places by continent. For example, Suriname is often overlooked when most people think about the Caribbean despite its similar colonial history and its membership in regional organisations such as the Caribbean Community (CARICOM). 8 Yet, there is absolutely no explicit indication of the nation's affiliation to the region in the TGN. Further, its index path World (facet) > South America (continent) > Suriname (nation) provides no possible way to collocate Suriname-indexed items with items of other Caribbean territories.
Like Suriname, Belize is often overlooked when people hear the name Caribbean, but it is another mainland country affiliated with Caribbean regional organisations. However, unlike Suriname, it has several index paths. Of those paths, at least one explicitly indicates some affiliation with the Caribbean and another one, branched under North and Central America (continent), would allow for a continental-wide collocation of Caribbean-related items since most island territories are directly or indirectly hierarchically linked to this continental heading. Perhaps most interesting is the TGN's use of a further term, Central America (general region), in one of Belize's hierarchies. This raises further uncertainty as to why the TGN has not implemented an umbrella term under North and Central America (continent) that physically encompasses, at the very least, all of the insular Caribbean. 9 By labelling 8 CARICOM (formed July 4, 1973) uses functional cooperation to achieve regional integration in the Caribbean, with four main driving pillars: economic integrity, foreign policy coordination, human and social development, and security. There are fifteen member states and five associate members. 9 The insular Caribbean is generally considered to be the islands extending from the northern coast of South America all the way up to and including the Bahamas and Turks and Caicos. If the TGN included a term such as Caribbean (general region) for this insular definition, it could also implement a term such as Circum-Caribbean (general region) that could account for politically associated mainland territories and neighbouring countries.  Caribbean territories under North and Central America (continent) but not having an intermediary term (or listing them under Central America [general region]), the TGN implies that Caribbean territories are part of North America, which is simply not the reality. 10 The final structural inconsistency that I will mention is that British West Indies (general region) is not indexed below a continent like other political entities. Meanwhile, the Netherlands Antilles (former nation/ state/empire) 11 and Antilles Françaises (general region) have been structured underneath a continental parent-although doing so poses unique challenges due to colonial histories. 12 While physically in the North and Central American area, politically the Netherlands Antilles is affiliated with the Netherlands, in Europe. Thus, the island group appears under both continental parents (see Figure 6). The same conundrum arises for overseas territories like Martinique. However, while confusing, these multiple continental headings are beneficial because they enable collocation from a geophysical or political perspective, and better reflect the complexity of some Caribbean places.

Ingrained North American and European Bias
An overwhelming number of Euro-American sources and contributors inform the place names and historical notes comprising the TGN, and their Western biases are inherently embedded in the system, as seen by the Eurocentric focus in historical notes on authority records and a lack of Indigenous place names included in each record's list of alternative names. Sources and contributors are included toward the bottom of each record's web page (Figure 7). A quick look usually confirms that most contributors are North American or European, like the Cambridge World Gazetteer, Webster's New Geographical Dictionary, and UN Terminology bulletins. A cursory glance at this section for more 10 At some point between September 2020 (when this original research concluded) and May 2021 (when this article was submitted), the TGN did just this, integrating a new physical feature index term Caribbean (general region). Time did not permit investigations of this newly added term and its impacts on the TGN's representation of the Caribbean before the article's publication. Further research is therefore required to determine how the addition of Caribbean (general region) has impacted the TGN's representation of the Caribbean. 11 It is also unclear why the TGN selectively applies the place type of former nation/state/empire since every territory in the Circum-Caribbean that is today independent constitutes part of a former state or empire. 12 Though not discussed previously, this challenge of trying to index places physically and politically leads to multiple index paths by way of continental hierarchy paths.  Additionally, Notes are decidedly biased to Eurocentric historical perspectives. Most notes indicate when territories were colonised, but very few include important national information such as dates of independence. Such significant dates should be considered unequivocally historic and given, at the bare minimum, equal importance as the colonial histories that presently dominate the Notes. Some notes falsely claim the extinction of Indigenous populations. 13 Further yet, there is a sense of contradiction that occurs between the Notes on some authority record pages, such as the one for Guyana. Despite making comments on Amerindians in Notes on other authority record pages, the Notes for Guyana simply state that "little is known" about the Amerindians. Even a cursory online search will show that much research has been conducted about these Indigenous populations.
Further, the vast majority of Caribbean-related preferred names in the TGN do not offer any Indigenous names in their Names list, going directly against the TGN's claim that it is focused on "the historical world" and "adding . . . Pre-Colombian place names" to parts of the world impacted by post-Colombian colonisation (Getty Research Institute 2019). In fact, during my research, the only Caribbean authority records in 13 There are still Indigenous groups living in several Caribbean territories, including Dominica, Guyana, and St. Vincent. the TGN I found with an Indigenous name included in their list of variant names was Jamaica (nation) and Jamaica (island), which both listed Xaimaca as an "Arawak" name meaning "land of the springs." However, even the use of the term Arawak to denote a language is vague because there are multiple distinct Arawak languages, as evidenced in the Diachronic Atlas of Comparative Linguistics. 14 These aspects demonstrate that the Getty still has a lot of work to do to make its vocabulary anti-colonial and more inclusive of the region's multiculturalism. The only way that these Euro-American biases will be overcome and balanced by more neutral and culturally relevant perspectives is if the Getty takes action to involve Caribbean institutions and experts on the region's social, political, and geographic histories and cultures.
Overall, the difficulty of defining and delimiting the Circum-Caribbean is reflected in the Getty TGN's attempts to organise this region into an indexable and hierarchical form. This analysis demonstrated that there are many weaknesses in how the TGN depicts the Circum-Caribbean region. The most notable weaknesses include a recommended hierarchy that often omits important place names or relationships that could enable additional discoverability and an overabundance of Euro-American sources and contributors that leads to a biased, Eurocentric perspective of this previously colonised and multicultural region. One of the redeeming qualities of the TGN is its attempt to recognise geography from both physical and political perspectives, although this brings its own set of challenges.

The Getty Thesaurus of Geographic Names® and the West Indian Postcard Collection at Cambridge University Library: A Case Study
This section summarises the methodology and findings of a case study in which I indexed the West Indian Postcard Collection (WIPC) at the RCS department at the UL using augmented applications of the TGN. 15

Background and Methodology
The level of cataloguing across different RCS collections varies widely. The WIPC had only undergone minimal cataloguing prior to my arrival. There was an eleven-page, typescript, hard-copy catalogue that listed postcards at item level with an accession number, the published caption, and sometimes a short description of a set of cards. I sought to augment the application of the TGN to improve the intellectual discoverability-determined by the number of TGN place names used as controlled access points-of this collection.
My primary practical tools in carrying out this research were two spreadsheets that I created: a consistency control spreadsheet and a catalogue spreadsheet. I began by listing each unique place identified in the WIPC postcards in one column in the consistency control spreadsheet, so that each row represented a unique place. Each place name was listed only once. I then added additional columns to analyse different ways of indexing using TGN names. These "listing" columns created a space for me to do the following for each place name: • List only the preferred names in the preferred hierarchy (see Table 1, row 1) • List the preferred names in preferred and additional hierarchies (see Table 1, row 2) • List the LOD names for preferred names in the preferred hierarchy (see Table 1, row 4) • List the LOD names for preferred names in preferred and additional hierarchies (see Table 1, row 5) Breaking down different applications of the TGN in these ways allowed me to see how many more controlled access points could be gained by incrementally incorporating preferred names in additional hierarchies and then also by including LOD names linked to preferred names.
I then inserted calculation columns to the right of each of the aforementioned listing columns. The calculation columns counted the number of names in each listing cell left-adjacent to the calculation cell. Finally, two additional columns used these counting columns to calculate the rate of improvement of accessibility. One column divided the number of all preferred names in all hierarchies for a given place by the number of preferred names only found in the preferred hierarchy for that place (see Table 1, row 3). The other column divided the number of all preferred names and their corresponding LOD names from all hierarchies for a given place by the number of preferred names only found in the preferred hierarchy for that place (see Table 1, row 6). The average of all counting columns was then calculated and inserted as a total row at the bottom of the consistency control spreadsheet.
The consistency control spreadsheet served as a tool to ensure my own consistency when indexing the WIPC in the catalogue spreadsheet. Primarily using only information available on the postcard itself, I listed each object as a row in the catalogue spreadsheet, including the accession number, maker(s), published title, and postcard set title (if applicable) in corresponding columns. I then indexed each postcard in the catalogue spreadsheet by identifying the location depicted in the postcard, finding that place's row in the consistency control spreadsheet, and copying and pasting the corresponding listing and calculation columns from that row into the corresponding object's row in the catalogue spreadsheet (ensuring that calculation formulas adapted to the new spreadsheet). Thus, the collection was indexed using multiple applications of the TGN.
Finally, the averages from each spreadsheet were averaged to give a final overall result (see Table 1). These are the rates discussed below. The calculations generated in the consistency control spreadsheet represent the truest average (in this project) for how much the intellectual discoverability (as defined above) of Circum-Caribbean materials could benefit from each indexing approach, since each place was listed only once in that sheet. Ideally, these same steps and calculations would be applied using a similar consistency control sheet which incorporates all Circum-Caribbean places identified in the Getty TGN (not just those places found in the WIPC).

Results and Observations
There were sixty-three unique places identified in the WIPC that could be identified in the Getty TGN. These sixty-three places are represented as sixty-three rows in the consistency control spreadsheet, and the 240 postcards that were catalogued are represented as 240 rows in the catalogue spreadsheet. Occasionally, I conducted minimal additional research to determine which parish the location of a postcard was in if the city itself was not listed in the TGN. This allowed the postcard to be indexed more specifically based on the available index terms in the TGN rather than just by its territory.
Since some territories are represented in the collection by multiple postcards, the calculations performed in the catalogue spreadsheet provide the specific rates of improvement of intellectual discoverability through expanded indexing specifically for the WIPC. Surprisingly, the calculated results in the catalogue spreadsheet did not differ greatly from those in the consistency control spreadsheet even though several places are represented substantially more than others in the WIPC (and therefore in the catalogue spreadsheet). What quickly became clear was that places benefit significantly from having a more extensive TGN authority record. In short, if there are no LOD names or only one name variant-for example, West Indies only has one-including the LOD really does not do much to improve the intellectual discoverability of that place. However, a place with dozens of LOD names-for example, Bermuda and some places in Jamaica-benefits significantly in terms of intellectual discoverability by having those LOD name variants included. In the context of this project, which considers a higher number of controlled names as an aid to intellectual discoverability, this statistical result demonstrates the importance of having additional LOD name variants listed on the authority record.
Furthermore, having numerous index paths (i.e., additional hierarchies) may significantly benefit the intellectual discoverability of a place because it means that there are more (regional and sub-regional) place names under which the (national/territorial) place can be found. This also means that users conducting broader research-for example, trying to identify all material pertaining to Leeward Islands rather than one specific island-could benefit from including additional names found only in the additional hierarchies. In many cases, broader place names (such as sub-regional groupings) are not included in the TGN's preferred hierarchies.
Based on observations that more LOD names and more available index paths lead to increased discoverability, the places that benefited the most from a broader application of the TGN thesaurus are Montego Bay (inhabited place), which gained an additional thirty-four access points using LOD, and Bermuda (dependent state), which gained an additional thirty-three access points. It is worth noting that several other places in Jamaica (Blue Mountains, Spanish Town, Westmoreland, Port Antonio) gained twenty-nine or thirty additional access points using LOD, likely due to the fact that they have longer index paths that include more sub-regional, country, parish, and county names as well as corresponding LOD names for each of those places. Places in Barbados benefited the least from these broader indexing applications. They gained only an additional eleven to thirteen access points through LOD because Barbados and its places do not have LOD names lists as long as other places like Jamaica and, unlike many places in the Caribbean, Barbados only falls under one sub-regional name in the TGN: British West Indies ( general region). Most places fall under a political or historical subset like this one as well as additional geophysical subsets like Lesser Antilles, Greater Antilles, Leeward Islands, or Windward Islands, which thereby increase index paths and consequently the number of place names, which in turn increase the number of LOD names. 16 It is interesting to note that the term West Indies (archipelago)-which would be useful for a user seeking a broader regional collocation of results-has only LOD name listed on its authority record. If there were more names listed, a broader application of the TGN could greatly improve accessibility through this term. For example, if terms like Caribbean were linked as LOD names for West Indies, users searching in a database with enabled LOD could retrieve results labelled as West Indies or Caribbean. It is unclear why this regional labelling has not received greater attention in the TGN, especially when there are several historical terms that could be added as LOD names to further expand discoverability. 17 The data in Table 1 shows that broadening the application of the TGN can significantly improve discoverability by creating more controlled access points (names), especially when systems employ LOD frameworks. The Getty recommends using the TGN's preferred names through the preferred hierarchy. These names and hierarchies are what the RCS department uses to index their collections since their catalogue database does not have the capacity to handle importing the TGN's LOD. The data above indicated that when all preferred names from the preferred hierarchy as well as additional hierarchies 16 For each preferred name (expressed as a row in the consistency control spreadsheet) and each postcard (expressed as a row in the catalogue spreadsheet), this rate of improvement was calculated by dividing the "Total No. of Preferred Names from All Hierarchies" by the "No. of Getty TGN's Preferred Names in Preferred Hierarchy." 17 The recent addition of Caribbean (general region) likely alleviates some of these concerns. The delimiting definitions of "West Indies" and "Caribbean" would need to be carefully considered and thoroughly investigated prior to conducting further research on the TGN's addition of Caribbean (general region) in order to determine which TGN places are linked to West Indies hierarchies, which are linked to Caribbean hierarchies, and which are linked to both. were included, the average increase in accessibility was just more than doubled-2.038 times the number of access points when compared to using only the TGN's preferred names in the preferred hierarchy in the catalogue spreadsheet and 2.112 times the number of access points as calculated in the consistency control spreadsheet (Table 1, row 3).
Intellectual discoverability, in terms of the number of access points, further increased when the use of the TGN's LOD names in the indexing approaches was considered. As seen in Table 1 (row 4), even including LOD names just for the preferred names in preferred hierarchies gave an additional 13.667 access points in the consistency control spreadsheet and 12.404 additional access points in the catalogue spreadsheet, an average increase of an additional thirteen controlled names between the two spreadsheets (rounded down to the nearest whole number). Finally, when the indexing approach was expanded to include all preferred names in all (preferred and additional) hierarchies and all of those names' relevant LOD names, accessibility increased yet again. On average, the number of access points increased from 3.746 (using only preferred names in the preferred hierarchy) to 28.650 (using all preferred names in all hierarchies and their corresponding LOD names) in the consistency control spreadsheet, an increase of 24.904 additional access points. In the catalogue spreadsheet, the number increased from 3.324 to 24.955, a difference of 21.631 access points. On average, then, the number of controlled access points between the spreadsheets increased by twenty-three when all preferred names and their LOD names in both preferred and additional hierarchies were included. Finally, by dividing the total number of all possible access points including preferred names in all hierarchies and their LOD names (Table 1, row 6) by the number of preferred names in preferred hierarchy (Table 1, row 1), the overall average discoverability was seen to increase by just over seven-fold.

Conclusion
Authority control seeks to standardise aspects of cataloguing and indexing. Controlled vocabularies and thesauri-a key component of authority control systems-might help in standardising the terminology used and in facilitating interoperability across software and institutions. However, they ultimately can hinder accurate indexing with regard to the Circum-Caribbean (and many other non-Euro-American regions) by leaving out valuable and nuanced access points. The controls we use to classify, categorise, and organise our collections in informational systems have become dangerously ingrained in the daily grind of processing and administering collections, to the point of becoming second-nature for veteran cataloguing professionals. This project and my ongoing research seek to invoke an "infrastructural inversion" (Bowker and Star 2000, 156) in which we think more critically about what it means to organise information and how the decisions made in creating information organisation systems can influence knowledge production, perpetrate biases, and affect information retrieval. Considering how information is organised and what knowledge is created or privileged in the construction of that organisation is essential to ensuring equitable access to all collection materials.
This research demonstrated that the intellectual discoverability of Circum-Caribbean materials in collections could be enhanced if a dual-indexing approach is used when indexing with the TGN. In this way, general (regional) place names and specific (territorial/national) names would both be included. As scholars such as Linnea Marshall (2003) have argued, broad and specific subject headings (or, in this case, geographical indexing) serve different sets of users, and it would be beneficial to consider how best to serve both sets of users. I have begun studying several other authority controls from a Circum-Caribbean perspective, but I have yet to find one that is nuanced enough to adapt to the peculiarities and Creolistic aspects of the Caribbean or that does a good job of representing the region's Indigenous histories and languages. In conclusion, while information standardisation has come a long way and certainly has its advantages, there is room for improvement. GLAM professionals could better serve their users by questioning the authority controls, controlled vocabularies, and cataloguing procedures they are implementing. Doing so could reveal the biases in the systems and cataloguing practices being used and could demonstrate how their use suppresses materials related to certain places and cultures. In the case of the TGN's representation of the Circum-Caribbean, a dual-indexing approach is one possible strategy to make Circum-Caribbean-related materials more visible.