THE TRANSNATIONALISM ROOTED IN PRODUCTION AND THE ‘PLACE’ OF THE GLOBAL UNION

Abstract In recent years, the idea of “new labor internationalism” has inspired controversy. The concrete developments emphasized by this debate (the emergence of “global unionism”) combined new strategies and traditional forms of union organization, confusing the established frameworks. In search of the nature of these hybrid arrangements, we explore what lies behind the “invention” of the global union as a new type of union organization, noting how the ideas of rooted transnationalism and network organization offered partial solutions to conflicts that permeate the relations between corporate and union powers. In particular, we discuss the emergence of global agreements and union networks in transnational corporations in manufacturing industries, and their adoption in Brazil. Our conclusion is that although the institutionalization of global unions has been hitherto precarious and contested, it has coherently articulated seemingly contradictory principles. However, given these hesitant compromises and tentative foundations, a critical and multiscalar sensibility is vital to unraveling the significance of the transnational practices of trade unions today.

a "rooted cosmopolitanism" of activists whose "most interesting characteristic is how they connect the local and the global" ( Tar row, 20 05: 2). Thus, the distinctive mark of the new contemporary internationalisms was the conception of a deep-rooted transnationalism inspired by networked organization.
Unions were not immune to these inclinations but incorporated them with some particularities. By and large, studies on new internationalisms relegated labour movements to a lesser status, since they were usually perceived as incompatible with the logic of networks (Castells, 1999: 424). In the early 2000s, while the internationalization of social movements had spurred extensive academic research, "international unionism continues to attract little interest outside labour activist circles" (Josselin, 2001: 169). Gradually, however, this situation was challenged by scholars who saw the emergence of a "new labour internationalism" as an important component of union renewal processes in the context of globalization, a thesis that has sparked important controversies (Brookes & McCallum, 2017). unionism requires specific treatment since its structures were built with the historical sedimentation of legally protected and institutionally recognized prerogatives in different spheres, most notably by national labour rights, but also on an international scale. Despite the recent developments in this field, what is conventionally called 'global unionism' is the direct descendant of the 'official' or 'institutional' international unionism established more than a century ago. As Peter Evans noted, the global trade union movement ar t iculates a complex set of local, nat ional, and inter nat ional power s, requiring the creative combination of hierarchical structures and networks (Evans, 2010).
In this article, we explore the 'invention' of the global union as a sui generis entit y that, unlike national unions under usual circumstances, cannot rely on state guarantees to exercise its prerogatives. We argue that global unions have resorted to the principles of deep-rooted transnationalism and networked organization to mitigate the limitations of this situation.
An important conclusion of the research, however, is that these principles were not always a goal in themselves. Rather, they represent the strategic adaptation of global unionism to a contested and shifting terrain, which led to introducing innovations that challenge traditional organizational logics while being closely associated with previously established str uctures of union representation.
Specifically, we investigate the development of global union policy on transnational corporations (TNCs) within IndustriALL Global Union -the larger global union dedicated to manufacturing sectors -and its predecessors in the chemical and metal industries (the ICEM and the IMF 2 , respectively). In this, a sociological approach focused on the elements of conf lict that underlie the 'social dialogue' discourses adopted by global unions proved to be fruitful, allowing us to identify, in the primary sources, the ways in which the ultimately antagonistic nature of the relationship between unions and companies influenced the development of global unionism, whose powers cur rently nav igate, to a great extent, the contested terrain created by union pressure on corporate governance procedures untouched by state supervision.
From this perspect ive, we analysed semi-str uctured inter v iews conducted with 52 interlocutors between 2014 and 2020. A first set of interviews covered promoters of global unionism as a broad union strategy to deal with TNCs. This batch included 4 advisors and 19 union officials (25 interviewees in total) distr ibuted as follows: 11 Industr i A LL , ICEM or IMF off icials; 5 officials in other international unions; 3 in foreign union organizations; 6 in union centrals and metal or chemical workers' confederations in Brazil.
The second group, formed by 27 interviewees, included workers, workplace representatives, and local union officials who participated in union networks promoted by IndustriALL in specific companies. Additionally, we analysed relevant documents, such as congressional resolutions, bulletins, pamphlets, 3-27 ARTICLE | RICARDO FRAMIL FILHO AND LEONARDO MELLO E SILvA and other items for external dissemination, as well as internal records provided by var ious union organizations, such as meeting minutes, activit y reports and communications. Union activities were also observed, from major Industr i A LL cong resses to smaller events, such as net work meet ings to address specific issues.
After brief ly contextualizing the history of sectoral or industrial union bodies at the international level, in the first part of the article we analyse how these organizations found in the role of TNCs' counterparts a path to reframe the relevance of their prerogatives. Based on the qualitative analysis of the material, focusing on the evidence on the global unionism promoters' concerns, we seek to characterize two main elements: the arguments that supported the conception of the global union as a project for transforming international unionism, on the one hand, and the resistance to this proposal, on the other. In the second part of the article, we rely on data about IndustriALL's activities and research evidence on the Brazilian case to assess how the global unionism model described by the research planted its roots.

THE "OLD" AND THE "NEW" TRADE UNION INTERNATIONALISM
When considering the theory produced on global unionism in recent years, the nature of the phenomenon is not always agreed upon. A first batch of scholarly research contrasted a diplomatic and bureaucratic internationalism with the network approach of engaged militants (Webster et al., 2008), whereas a second approach prioritized the investigation of new forms of transnational labour regulation through the lens of Industrial Relations (Papadakis, 2011).
Nevertheless, the convergence regarding the practical experiences associated with new labour internationalism was remarkable. Notably, organizations that until recently had operated in relative obscurity were deemed central: the industrial or sectoral internationals, union bodies responsible for representing specific industries or professions at the international level, known today as Global Union Federations (GUFs). As McCallum (2013: 5)  workers' categor ies were also created (Stevis, 2020). Or ig inally known as Inter nat ional T rade Secretar iats (ITSs), they were the GUFs precursors.
In general, trade secretariats were seen as subordinate appendages, bodies whose relevance was constrained by the 'mundane' concerns of specif ic g roups. In the 1920s, for example, Lor win (1929: 99) considered that "the work of the Secretariats was narrow in scope and modest in results," whereas a report for the US government concluded that the resolutions of the metalworkers' international was devoid of "general interest" since they were limited to discussing "purely trade-union tactics and methods, which include the prevention of the importation of strike breakers, travel benefits, maintenance of a defense fund, and exchange of union data" (U.S. Department of Labor, 1920: 77).
The kind of union strategy discussed in this article was first glimpsed in the second half of the centur y, when the emergence of multinational companies opened a new area of activity for trade secretariats. Since the 1960s, 'world company councils' started to appear, union bodies designed to coordinate collective bargaining within the same company across multiple countr ies (Levinson, 1972). These councils were established in dozens of multinationals, notably in the chemical and metal industries 5 , but they fell short of the expectations of their creators who, guided by an "evolutionary optimism," would have seen in the collective bargaining internationalization an inevitable consequence of the production internationalization, a prediction that was never fully realized (Ramsay, 1999).
Nevertheless, as heirs of this experience, ITSs were well positioned to respond to what emerged as a central trade union concern in the context of globalization -the intensification of the production transnationalisation propelled by transnational corporations (TNCs). In this context, the sectoral internationals sought to strengthen their structures, accelerating the historical trend of reducing the number of organizations accompanied by expanding the representation basis of individual internationals (Bourque & Hennebert, 2011). Today, the largest GUFs each cover broad sectors of economic activity, such as public ser v ices, pr ivate ser v ices and manufactur ing. In 20 02, the name 'trade secretar iats' was abandoned in favour of 'Global Union Federations' and, more recently, some GUFs started using the designation 'global union' 6 .
IndustriALL Global Union is a typical example of this process. Able to trace its origins to organizations founded more than a century ago, it adopted its current form in 2012 and claims to speak on behalf of over 50 million workers in 140 countries, distributed throughout multiple branches of industrial manufacturing. It resulted from a merger between three important GUFs: Although global agreements can be compared to collective bargaining, analysts have highlighted the particularities of these instruments, which are subordinated to a sphere of private governance 8 surrounded by uncertainties.
The most common interpretation defines GFAs as a "union response to the social responsibility movement of corporations" (Hennebert, 2017: 117) or, in stricter terms, as a direct evolution of 'codes of conduct', charters of principles published by large companies in response to international campaigns protesting labour, social, and environmental violations. What separates agreements from codes is that the former result from negotiation processes in which unions are recognized as leg itimate signator y parties. Nevertheless, GFAs share important features with unilateral codes: they celebrate general principles (a 'framework') that, in most cases, have indeterminate temporal validity; are voluntary, that is, companies are not under an obligation to participate or to maintain this participation; finally, they are not directly enforceable, that is, they lack state guarantees for compliance with their provisions.
That global trade unionism pr iv ileged this sphere has fr ustrated certain expectations, not least due to GFAs being vulnerable to the criticism that they reinforce the privatization of international labour law (Thomas, 2011). Faced with this suspicion, much of the research on the subject has focused on searching for concrete evidence of the practical use of GFAs, which generally suggests conclusions that are difficult to generalize (Krause, 2018). In this article, we present an alternative and complementary perspective.
We arg ue that, notwithstanding the obvious relevance of evaluating the normative effectiveness of GFAs, a unilateral focus on this dimension may lose sight of the role played by codes of conduct in legitimizing the role to which the GUFs aspire. An important aspect is that codes of conduct implied the unprecedented recognition by powerful companies that their operations could be subjected to global rules. Presenting themselves as the workers' voice in the definition of these rules, GUFs emerged as candidates to embody a global union mandate that was previously absent. From the perspective of sectoral internationals that were traditionally relegated to a secondary role, the signing of global agreements was a momentous milestone that signalled the feasibility of this ambition. Thus, even when assuming apparently innocuous forms, GFAs boost the consolidation of global unionism in a broader sense.
However, with formal recognition alone, global unions are still grasping at straws. Aware of the limitations that had hindered previous experiences, the GUFs proposed a profound reconsideration of the nature of international unionism and, thus, found in deep-rooted transnationalism a path to re-signify their sectoral vocation and defend the need for a new type of organizationthe global union.

The global union in manufacturing industries
Among the federat ions that would create Industr i A LL , it was the ICEM that first conceived the essential components of a global union in manufactur ing sectors. In its f irst cong ress, in 1995, the organization arg ued that corporate power had eroded union inf luence over important decision--making processes, which could no longer be super vised by governments or parliaments, but were conducted "behind the boardroom doors of the major mult inat ional companies" (ICEM, 1996: 54). In pract ical ter ms, it demanded a "f irmer def inition" of the responsibilities associated with each of the two types of international union bodies (the secretariats and the "umbrella" federations), arg uing that "it is sometimes forgotten that the inter nat ional industr ial federat ions are pr imar ily organisat ions of industrial action, in fact industrial struggle, rather than political lobbyists" (ICEM, 1996: 55).
Thus, a pioneer ing for mulat ion of the aff init y bet ween industr ial vocation and deep-rooted transnationalism appears. According to ICEM, sectoral internationals should be concerned "with building practical links between workers; with actual pay and conditions; with supporting actual industrial disputes; with the application of occupational safety and health protections in individual industries and global companies" (ICEM, 1996: 60).
This argument was f undamental for reconsider ing the relevance of the GUFs' sectoral attr ibutions. Traditionally, the 'minor' day-to-day concerns of local unions had been out of the scope of a 'diplomatic' internationalism interested in more directly political and ideological issues or dedicated to lobbying in multilateral organizations. Through the lens of rooted transnationalism, however, these prerogatives started to be seen as an alternative to the 'top-down' approach of union diplomacy, a way of building initiatives 7-27 ARTICLE | RICARDO FRAMIL FILHO AND LEONARDO MELLO E SILvA "rooted in the day-to-day realities of members, and not the musings of remote international leaders" (Fairbrother & Hammer, 2005: 422).
In this model, deep-rooted transnationalism guides the creation of global union alliances that must be able to engage companies directly, which would justify, in turn, strengthening the GUFs as 'union counterparts'. It was no accident that ICEM mentioned global agreements as an "example of what can be achieved" (ICEM, 1996: 58). Platzer et al. (2011: 195) noted that in this period "the field of company-related activity has seen a qualitative strengthening in the functional profile of the federation," which they associated with the development of a "strateg ic perspective" aimed at building the necessar y conditions to the exercise of "the genuine function of a 'trade union'," that is, "to become an actor at the global-transnational level capable to 'negotiate and agree'" (Platzer et al., 2011: 191). The later use of the name 'global union', therefore, was not incidental or without purpose. This trend has been followed by other GUFs, but it has not gone unchallenged. The metalworkers' international decided to support codes of conduct in 1997, and, according to the organization, "it was clear from the outset that such codes were to be negotiated and would become agreements between the IMF and transnational corporations" (IMF, 2006: 2). A few years later, however, the federation would notice that companies were reluctant to accept it as a signatory party, a role they preferred to grant to unions in their home countries. The IMF's recommendations, consequently, were insistent on the federation's role in global agreements: "IMF should be involved from the start;" "An IMF officer or designee must sign the IFA [International Framework Agreement];" the agreement "should be negotiated by IMF and management at global level" (IMF, 2006: 2). These determinations confirm the hypothesis that GFAs were important to legitimize the position of global unions. In the words of the general secretary of the IMF at that time, global agreements would serve to "create an IMF identity" (IMF, 2003: 18). Along the same lines, when asked about the importance of these agreements, a director of IndustriALL stated that "in the first place, it guarantees the recognition of unions at a global level, to the Global Union Federations at large" (interview, 2014).
Today, the role of the GUFs is well accepted among unions engaged in global unionism, which is the subject of the next section. Corporate resistance, however, is more persistent. The International Organization of Employers, which presents itself as "the global voice of business," states it "takes no formal position for or against" GFAs, but warns affiliates that "this is not a Corporate Social Responsibility initiative" and that, even if their "legal status is untested," global agreements should be treated as contracts "of a legal nature" (IOE, 2013). On the other hand, the organization acknowledges that companies, unlike trade unions, prefer to see GFAs "principally as a mechanism for deepening dialogue, first and foremost, and not as an industrial relations exercise" (IOE, 2007: 8). When it comes to analysing concrete labour conf licts within specific companies, considering both corporate strategies and the organization of production is necessary. Ramalho & Santos (2018: 25) highlighted the complexity of the interactions between workers' agency and the transformations in global production networks, proposing "a relational understanding of the conformation of networked corporate strategies and of a deeply stratified working class, resulting in more complex conditions of consent and resistance." Jurav ich (20 07), in tur n, detailed how unions can use research to obtain information about corporate structures and use them strategically. This was an allur ing prospect for global unions and ICEM proposed that building "overall knowledge of the company's production, strategies, industrial relations policies and collective bargaining" would offer "a convincing demonstration that an industrial trade union international is not a remote bureaucracy, but an essential provider of data and information for concrete, day-to-day union work" (ICEM, 2005: 56).
The GFAs must be understood within this context. The fundamental thing is that union incursions into the field of "corporate governance," across different scales, suggest an attempt to establish some degree of public regulation over spaces that escape democratic oversight. Therefore, this always involves a dimension of conf lict, and Stevis (2009) was correct when he defined GFAs as a "battleground" between unions and companies. This contested nature is precisely what allows us to reconsider the significance of the mandate claimed by the GUFs, which cannot be reduced to a rearrangement of representation powers previously held by other union bodies.
Usually, a company is expected to carr y out truly global strategies, but public reg ulat ions, and unions, are constrained to local or nat ional jurisdictions. Unlike other movements, to which the "global" may appear as a diffuse phenomenon, in the case of unions facing TNCs, corporate decision centres correspond to global authorities that can be targeted directly. This is why deep-rooted transnationalism can articulate pr inciples that, from a tradit ional perspect ive, would be seen as incompat ible, namely, local union demands and international organization.
The old world company councils proposed something similar, but their weaknesses, and the recent transformations in the organization of production, made global unions seek more f luid ar rangements, capable of matching cor porat ions net worked organizat ion. In this regard, it is relevant that large companies offer not only global counterparts, but also a cartography for global union alliances. In essence, the idea that there is a "latent union strength" (ICEM, 1996: 56) embedded in the global production disposition qualif ies the meaning attributed to GFAs. As with universal declarations of rights, these agreements establish an abstract global jurisdiction, but, unlike them, they also demarcate specif ic const ituencies -the workers under the authority of the same company in determinate localities, a group 9-27 ARTICLE | RICARDO FRAMIL FILHO AND LEONARDO MELLO E SILvA that could, potentially, organize itself and make itself represented at the global level.
From this perspective, the GFAs limitations stem not from union naivety regarding codes of conduct or the corporate social responsibility shortcomings, but from the fact that their formal scope is not always accompanied by a strong enough organizational basis 9 . In the words of a Brazilian metalworkers' union official, "the limit of the Framework Agreement is that it will only have life, effectiveness, with union action supporting it; without a union defending that, the company will not comply" (interview, 2014). Another interviewee, from the chemical sector, considered that "for the agreement to be functional, it must have a union network alongside it. Which should come first, the network or the agreement? The network. It will raise issues, demand that the agreement has certain features" (interview, 2014). It is no coincidence that IndustriALL referred to global agreements and union networks, in tandem, as its "Way Forward for the Future" (IndustriALL, 2012b).

UNION STRUCTURES AND NETWORK ORGANIZATION
Deep-rooted transnationalism suggested that international unionism "may be refig ured away from vertical representation toward the coordination and management of a network of trade unionists stretching across the global economy" (Wills, 1998: 127). Such concerns are relevant to global unionists, but considering the position occupied by the GUFs in representation structures is impor tant since, even when the name 'global union' is adopted, these entities cannot discard federative str uctures completely. After all, GUFs are financed by national unions that define their priorities and elect their officials. Direct recruitment of workers is possible, but rare, usually limited to situations where union representation in not established in a territor y.
Not only would the resources available to the GUFs be insuff icient to support sustained efforts in this direction, but local unions would likely protest the usurpation of their attributions. As summarized by Garver et al. (2007: 239), "GUFs, which have affiliated unions rather than direct members, must rely on those affiliates for democratic legitimization and for organizing local actions." Global unionism, therefore, has long been operating as a cooperat ion bet ween diversed union organizat ions, an approach that favoured adopting the idea of networked organization.

Union networks in manufacturing industries
The ICEM's foundational propositions emphasized the need for union action to be "planned on an international basis right from the start" (ICEM, 1996: 55), which could suggest a pr ior it y to the organization's global mandate.
In reality, however, its resolutions conveyed a more realistic compromise. The international recognized that "the power of any union structure resides in its local branches," criticised a "centralising view" and argued that union strategies should be carried out as "a series of networks" (ICEM, 1996: 56).
Guided by its broader goals, ICEM focused on the creat ion of 'company networks', structures that should operate with the permanent communication between the international and 'union correspondents' in the factories of a single company in multiple localities. The contours of such networks were not precisely defined at first, but it was suggested from the beginning that these str uctures should not be limited to information exchange, that is, they should also support joint initiatives, enabling globally unif ied strategies to face TNCs.
In the second ICEM congress, in 1999, the organization's leadership assessed the networks that had been created up to that point in a positive light and determined that the "exchange of strategic information links previously separate g roups of workers," which would have provided "a f irm basis for solidarity" (ICEM, 1999). About ten years later, the IMF had established general recommendations that, if not to be read as a rig id model, illustrate what should be expected from networks in traditional manufacturing industries. Eager to set up alliances capable of negotiating with TNCs, the international highlighted the need to reach the "real decision-makers within the enterprises" and proposed to recognise the "political mandate" claimed by networks that would, therefore, speak on behalf of participating unions when addressing the company (IMF, 2010: 2). The compromise was the definition that "each union itself decides over who participates or not in the net works," although it was added that, in doing so, they should recognize "the importance of representation and participation from the shop f loor" (IMF, 2010: 3). Likewise, it was agreed that unions would encourage the participation of representatives from the workplace, but that was followed by the guarantee that this would happen "according to the practices in the respective countries" (IMF, 2010: 4).
In essence, what this means is that local arrangements should not be IndustriALL reaffirmed these principles in its "Charter of Solidarity," a document that reg ulates the aff iliated unions' responsibilities and the global union's "mandate" regarding, among other things, union networks and global ag reements. Deep-rooted transnationalism is celebrated with the provision that "affiliated unions will seek to ensure that member activists and representatives are fully involved" (IndustriALL, 2012c), but the ambiguous wording to refer to the participants of this movement (member activists, which in the Spanish version is indicatively translated simply as sindicalistas, i.e. 'unionists') highlights the tension that permeates the exercise of a mandate that depends on the voluntary cooperation of autonomous entities. This approach is guided, on the one hand, by the ambition of building translocal connections capable of supporting a global policy to confront TNCs and, on the other, by the need to disarm conf licts with local unions, that is, by peaceful coexistence with established unions' powers, who might see the emergence of global unions as a threat. In this process, global unionism is both benef ited and constrained by integ ration with pre-existing union structures.

GLOBAL UNIONISM IN PRACTICE
The previous sections identified the fundamental elements of a global unionism project whose apparently contradictory provisions reveal their coherence in negotiated arrangements that made the conception of the global union possible in the absence of state guarantees to the exercise of a representative mandate. In practice, the solutions presented so far mitigate tensions, but do not eliminate them completely, which means that institutionalizing the global union is still a hesitant and contested process.
Regarding relations with TNCs, the clearest evidence of the limits of this compromise is that most GFAs were established in companies headquartered in Europe 10 . In the case of IndustriALL, only 4 agreements out of 46 were sig ned by companies based outside the European Union and the United Kingdom (this group represents 91% of the GFAs signed by the organization).

Aware of this bias, IndustriALL sought to boost initiatives in other regions.
Analysis of the organization's action plans reveals a more balanced picture that is closer to the regional distribution of the headquarters of large TNCs in the studied sectors 11 , with targeted companies coming from Europe (63% of prioritized companies, compared to 59% among large companies in relevant industries), the United States (17% compared to 20%) and Asia (10% to 19%).
Also note the activit y in companies of Latin Amer ican (6%) and Afr ican Despite these efforts, regional imbalances in global trade unionism also result from the historical importance of European unions within the GUFs, whose federative structures are vulnerable to national inf luences.
About ICEM, Cumbers et al. (2008: 376) arg ued that the organization was not "a coherent movement as such but rather a coalition of diverse labour interests, different iated on spat ial, ideolog ical and sectoral g rounds." This situation was detailed in the years following its foundational congress by an important leader in the federation, who reported that "in the 1990s, the attempt was to make the networks work from plant to plant, despite the fear, if we can put it this way, from national union organizations. That was always the tension, that was the real problem" (inter view, 2020).
Initially, the international's leadership used networks to defy the union bureaucracies' inf luence more directly, which was facilitated by the inf lux of new potential allies (unions and workers in Eastern Europe and the Americas, for example). According to the interviewee, however, the conf lict deepened when ICEM explored building networks with a higher degree of autonomy, seeking alternative means of funding, and promoting spaces of articulation beyond the super vision of federative str uctures, a step that was deemed unacceptable by members that were critical of the organization's early policies.
After its second congress, ICEM's proposals were moderated, which prevented a break with important financial supporters.
If replacing them entirely is impossible, an alternative could be democratizing federative structures. IndustriALL has set itself the goal of promoting greater participation by unions from the Global South and permanent offices have been set up in Ur ug uay, South Afr ica, Russia, India, and Singapore.
The organization's second cong ress took place in Rio de Janeiro, in 2016, when a Brazilian metalworker was elected to the position of General Secretary.
The third congress was planned to take place in Cape Town (South Africa), but it was carried out in a virtual environment due to the COvID-19 pandemic.
However, since the GUFs do not directly control networks, adjusting federative powers may not be enough to counterbalance the inf luence of unions with greater financial resources, which often fund global initiatives directly (Cotton & Gumbrell-McCormick, 2012). Without losing sight of these diff iculties, the study of union networks in Brazil reveals that the global unionism model described so far allows even apparently weak local unions to inf luence the practical implementation of global policies.

Deep-rooted global unionism in Brazil
In a comparat ive study, Galhera (2016)  However, since global unionism depends on the voluntary cooperation of local representatives, the latter preserve the capacity or, at least, the legitimacy to represent workers in their jurisdictions, which means that they have been able to negotiate the terms of their incorporation into global initiatives that, if considered unfair or ineffective, could simply be ignored or abandoned, which would close the networks' access to relevant territories.
A German trade unionist interviewed for the research, for example, lamented that Brazilian allies usually send a different representative to each international meeting, making it difficult for the networks to operate consistently (interview, 2014). From the local perspective, however, rotating representatives is important to preserve good relations between unions, and thus the practice has been maintained. This situation reveals an unforeseen consequence of the dependence on established union powers -local structures may constrain networks, sometimes controversially, but they can also protect global initiatives from capture by powerful national funders, which favours the GUFs' mediating role. the GUFs offer purportedly democratic instances that can be questioned.
During a meeting of chemical unionists in São Paulo in 2016, for example, a local representative cr iticized the negotiations of a global ag reement, centralized in Europe, for ignoring local demands. At the time, the leader stated that, if IndustriALL did not address their concerns, the matter would be taken to the press. This situation shows that there are disagreements, but also that local unions see global unionism as a realm that should be claimed and contested, not ignored.
The articulation of local unions in Brazil within a network dealing with a chemical company headquartered in the Netherlands illustrates how these trends manifest themselves in practice. A Brazilian trade unionist noted that the initiative came from the Dutch trade union federation, "a project from the [unions in the] Netherlands, which wanted to know the regional particularities of the parent company, not only in Brazil, but also in other places" (interview, 2018). Meetings between local representatives from different factories in Brazil began to be organized in 2001. At the same time, arrangements were made so that local off icials could travel to the company's global headquarters.
The interviewee reported that, at that time, he held the position of "factory safety commissioner" 14 and did not know the factories operated by the company in Brazil besides his own: "The first time, I ended up using the network. I was new, I didn't know better. So, we presented the demands of [the factory in] São Bernardo to the global CEO" (interview, 2014). As contacts grew stronger, however, local priorities gave way to unified campaigns: "Knowing the realities and the inequality that existed [in the company] in the country, involving the distribution of bonuses, wages, benefits, we thought… 'so, how can a company wearing the same shirt have this difference in benefits?'… We started to create almost a single agenda" (interview, 2018).
As in the case of similar networks, national criteria were considered more realistic in formulating concrete demands, but international contacts were essential. The interviewee valued the support of the Dutch who, according to him, "always helped a lot," particularly in exerting pressure on the company's central management. He highlighted that solidarity also f lowed in the opposite direction. When Dutch workers were fired from the company in 2008, Brazilians stopped local production in protest: "We stopped for two hours at the units in Mauá, São Bernardo, and Recife… that had a lot of repercussions.
What does the Netherlands have to do with Brazil here? They are workers, we are in solidarity with them too. They are in difficulty, but they wear the same shirt" (interview, 2018). In this sense, the existence of territorially separated workers 'wearing the same shirt', that is, linked to the same company, is the direct justification for the existence of networks.  2012 and2013;and Colombia, in 2016. In this process, trade unionists from different locations pressured the company into taking part in annual 'social dialogue rounds', bargaining tables that occurred at the national and, on some occasions, regional level. Note that, although the company in question has refused to sign a global agreement, these spaces operate with a logic that is very similar to the one behind GFAs. Unions frame the idea of a 'dialogue' in terms of collective bargaining and mobilize workers to demand compliance and the progressive institutionalization of what they perceive as negotiated deals, whereas the company prefers to limit these practices to an exercise of social responsibility and sharing of ideas. This indicates that the global union policy that support GFAs can produce arrangements that, despite being much less visible to outsiders, can be just as relevant to the practical consolidation of global unionism.
Finally, the contact of local unionists with global unionism opens up the possibility of establishing alliances beyond company-based cooperation.
These processes are intertwined, which reinforces the suggestion that global unionism can inf luence union practices in a broader sense, even when its role is not immediately perceptible. In the words of a trade unionist who was part of a union network in another European company in the chemical sector: IndustriALL holds a [company network] meeting in Denmark. Since we are there, we talk to these unions and propose: we want to get to know you, how the union works. We also want them to come to Brazil and we close an agreement. We have an agreement with Turkey that we signed in 2009 at an IndustriALL meeting -then ICEM -and it's going ver y well. We went there and they came here several times… With the French too. Recently, the [company] got mad at us. As we gain agility with international information, and in the case of the [company] it has European origin… [if ] an accident happens here, you put it in an email and everyone knows about it. There the relationship is different. Here, to get to the industrial manager, one goes through a lot of people. There, no, they have a direct relationship if someone needs to make a complaint. Quite often, thanks to what we release to them, from there comes a message to the factory, and the factory [manager] gets upset. (interview, 2014) These experiences demonstrate that what at first glance may appear as local or binational events are in fact intimately linked to the emergence of global unionism. The global union does not control these initiatives but enables a practice that is at once transnational, rooted, and multiscalar.
Seemingly trivial local demands, such as an incidental dissatisfaction of workers in a given factory, fuel a strateg y that unfolds across multiple scales: foreign unions' inf luence over global executives is mobilized to challenge the despotism of local bosses; translocal contacts favour the emergence of joint demands and campaigns; organization in the workplace, whose greatest weapon is the strike, is used to boost national and transnational campaigns. continuously. In the words of a metalworker unionist, "online, we exchange information simultaneously. It's right by the machinery, you can hear the guy cutting metal sheet there" (interview, 2014). In this way, deep-rooted transnationalism moves away from the kind of international trade unionism that is located "at the peak of a pyramidal structure several removes -and gatekeepersaway from any f lesh-and-blood workers" (Waterman, 2001: 315).

CONCLUSION: THE "PLACE" OF THE GLOBAL UNION
In the midst of the internationalist enthusiasm that marked the emergence of global unions, Richard Hyman considered that international unionism continued to be constituted by national organizations and questioned whether it would be an "extension of national experience, or whether in key respects it can be regarded as a distinctive social phenomenon." He suggested that we inquire how far the 'balance' between international and national trade unionism had been shifted; whether new forms of action were created or whether they would be 'borrowed' from national experience; if there is an 'ideology' and a 'discourse' that are "not necessarily rooted in national trade union practice" (Hyman, 2005: 138). Our results reveal that these questions could be turned on their heads. In the cases described in this article, in the end, global unionism, if sometimes begrudgingly, asserts its relevance not by its separation, but, on the contrar y, due to its ability to closely connect with local unions, that is, by the local rooting of global policies.
In fact, the dichotomy between the local and the global and the attempts to overcome it were prominent in the theoretical debates around new labour internationalism. Michael Burawoy (2010), in what was the best-known critique of this idea, accurately identified conceptual difficulties that continue to challenge researchers, but also advanced an essentially empir ical arg ument.
According to him, when one looks at the concrete experiences that would link workers to this new internationalism, "there's simply no there there," an expression that, with a polysemic trick, suggests the absence of a real place occupying the space opened up by imagination. Essentially, struggles would still be tied to the local, refuting the internationalist mirage. The relevance of Burawoy's broader arguments notwithstanding, the strongest counterpoints to his critique emphasized the permeability between scales and levels of action, which would disavow a dichotomous opposition. Along these lines, Lambert (2010: 389) argued that Burawoy "misses the significance of a new conception of globalizing the local, namely, its construction through a networked linkage between local places across geographic space." When they are separated from the underlying developments that support them, such statements can sound overly abstract or even enigmatic.
Our research, however, agrees with the conclusion that this is not an exclusively theoretical issue. In fact, in the case of global unionism, the challenge 17-27 ARTICLE | RICARDO FRAMIL FILHO AND LEONARDO MELLO E SILvA of devising strategies capable of crossing local, national, and global jurisdictions is presented in a very practical manner. It shouldn't be surprising, then, that the experiences presented in this article have often confused unionists themselves, who are accustomed to navigating well-defined hierarchies, authorities, and territorial jurisdictions. What is the place of networks in trade union structures? In the words of an IndustriALL director, "when we talk about networks, they don't fit well with our traditional union structures.
We have local unions, reg ional unions, union delegates, union off ices… Now we are talking about another structure that is not located somewhere.
We want to include everyone. Local people, regional people, national people…" (interview, 2014).
This approach poses important challenges to research on the transnational practices of current trade unionism. Above all, multiscalarity must also be an analytical sensibilit y. This becomes clear, for instance, when researchers are faced with situations in which, to defuse conf licts with established unions, GUFs deliberately step out of the spotlight. When this happens, the network and the global union seem to 'disappear' amid translocal contacts, and their policies maybe be described as mere tools at the disposal of established unions. As another Industr iALL leader arg ued, "the network should not compete with the traditional union. The network is a tool that the union has to link different factories. This is how the union should see the network: as the sound-car, as the website…" (interview, 2014).
This may seem deeply under whelming to the untrained eye, and in this article we have ventured to argue otherwise.
We have acknowledged, on the other hand, that the dependence on existing structures imposes significant constraints, a crucial point when it comes to the ability to reach different categories of workers, especially those that cur rently lack local representat ion. Thus, Cotton & Croucher (2009: 119) advocated abandoning the emphasis on collective bargaining, a policy that would prioritize established categories, but championed the "political decision" to dedicate greater resources to global unions, "the only institutions that can develop the collective exper ience, articulation and collaborat ion bet ween unions in the ways demanded by globalisat ion." van der Linden (2016) agreed with criticising strategies focused on collective bargaining and considered that the global unions would be better prepared to organize broad g roups of workers, suggesting that they should bypass outdated local bureaucracies.
Our conclusions indicate that the importance of collective bargaining for global unions should not be so easily dismissed. It was precisely due to the GUFs assuming a role in the relationship between the "us" of the workers and the "them" of TNCs that deep-rooted transnationalism could emerge as an alternative to diplomatic federalism, and collective bargaining was fundamental in this regard. Nevertheless, we can concede that, having secured pre-existing union powers, initially an imposition of the circumstances, is increasingly perceived as a virtue. Global unions seem most promising not when they attempt to assert their authority over a global mandate, although that might be beneficial at times, but when they focus on their expertise as weavers of union strategies across multiple scales, becoming "intermediaries or facilitators in multi-scalar initiatives" (Ford & Gillan, 2015: 14). 9 This is the position of the leader responsible for signing the first global agreement, which occurred in the food sector, who criticized the association of GFAs with corporate social responsibility and argued that the "original intention" of the policy was based on the collective bargaining strateg y developed by the International Trade Secretariats (Gallin, 2008). 14 Cipeiro, a worker elected to the Internal Commission for the Prevention of Accidents (CIPA).