Everywhere and nowhere at once: the challenges of following in multi-sited ethnography

Purpose – In multi-sited ethnography, “ following ” (of, e.g. persons, objects and events) is used as a device to structure fieldwork. The purpose of this paper is to problematize and substantiate the notion of following, illustrating that, when adopting a “ following ” strategy, the endless number of potential trails one could follow may lead a fieldworker to be both everywhere and nowhere at once. Design/methodology/approach – This paper is based on the experiences and insights derived from a multi-sitedethnographyofthestrategiccollaborationsthatemergedaftertheDutchhealthcarereformof2015.Fieldwork wasconductedbetween2015and2017,andconsistedofparticipantobservations,shadowingandinterviews. Findings – An approach well suited to studying the contemporary problems that cut across organizational boundaries, multi-sited ethnography is both valuable and more challenging due to: (1) the continuous need to negotiate access, which stimulates the researcher to reflect on his or her positionality in the field; (2) the inevitable pressure it puts on a researcher to “ unfollow ” their field(s) and to regain critical distance and (3) its perplexing ability to highlight the lack of a whole, unveiling instead a plethora of perspectives across sites which may or may not align. Research limitations/implications – This paper ends with three key considerations for future multi-sited research endeavours. Originality/value – Although the metaphor of following can help to structure fieldwork, the practice of following in multi-sited ethnography is not as straightforward as it appears: there are countless potential “ paths ” to follow, and researchers themselves must decide which trails to choose and when to step back and “ unfollow ” their field(s).


Introduction
In 2015, the Netherlands underwent a huge transformation in domestic governance: the decentralization of the organization of healthcare and its financing, which transferred responsibility from the central government to local municipalities and healthcare insurers. While local actors were still expected to deliver the same quality of healthcare (if not better), this transition was accompanied by significant budget cuts. As such, local actors were expected to transform their current practices into an "integrated" system, with a strong emphasis on (regional) collaboration. This collaboration was supposed to manifest in two forms: in a "narrow" collaboration between the municipalities and healthcare insurers (the financiers), and in a "broad" collaboration that also included other actors throughout the domestic healthcare system. Naturally, the need for such changes forced organizational actors, who often had no common history of collaboration, to suddenly and jointly transform the healthcare system into one that was both cost-efficient and tailored to providing quality, individualized care. Perhaps most challenging were the dissimilarities between the organizations that were expected to collaborate: the "commercial" healthcare insurers and the "bureaucratic" municipalities were not only organized very differently, they also had potentially conflicting goals and interests. My research focussed on the question of howand ifthe local actors managed to jointly construct the new, local forms of governance required by the healthcare reform.
Given the mutual sensemaking processes shared by actors across the healthcare system, multi-sited ethnography seemed to be a perfect approach. Multi-sited ethnography was first coined by Marcus in 1995. He described it is as the next step for ethnographya fitting approach to a rapidly changing society and, accordingly, evermore complicated research objects that do not confine themselves to a single location (Marcus, 1995). This approach turns ethnographic fieldwork into a matter of "being there . . ., and there . . ., and there!" (Hannerz, 2003). One way of pursuing a multi-sited ethnography is through what Marcus described as "following", for which he proposed six modes to help structure fieldwork. Following a certain object or subject allows researchers to naturally move from one site to another as developments unfold. By using a multi-sited approach in my own fieldwork, I was able to follow collaboration across the reformed areas of the healthcare system and understand how the various actors tried to construct new forms of governance both collectively and within the boundaries of their own fields. This practice of "following", however, turned out to be problematic. Not only were the boundaries of the field(s) themselves difficult to define but also the act of "following" turned out to be less straightforward than I had initially assumed. Rather than "naturally" being where I "needed to be", I sometimes felt as if I was everywhere and nowhere at once.
In this paper, I problematize and substantiate the notion of following based on my own experiences as a multi-sited ethnographer as I followed collaborative efforts across the Dutch healthcare system. To elaborate on my approach, I will begin by discussing the literature on (multi-sited) ethnography. After that, I will share three challenges I encountered in my attempts to follow collaboration across multiple fields in order to reveal how these challenges also turned out to be lessons; namely how: (1) the constant need to negotiate access to different field(s) stimulated me to reflect on my own positionality in the field; (2) the overwhelming number of potential trails to follow forced me to "unfollow" the field and to regain critical distance and (3) the patchwork of perspectives I observed throughout my fieldwork sensitized me to the chaos that not only I but also the local actors themselves had experienced. In a humbling way, multi-sited ethnography thus demonstrates the unattainability of the ethnographer's ambition to offer a holistic account. In conclusion, I will discuss both the notion of following itself and the implications of my research for future multi-sited-ethnographic "following" studies.
Multi-sited ethnography: fuzzy fields and puzzling problems Throughout my research, multi-sited ethnography proved to be a bittersweet experience. It was, simultaneously, both the best approach to studying my research topic and, given its many challenges, the worst. Multi-sited ethnographic studies that focus on interorganizational dynamics are still quite rare (Zilber, 2014). This is unfortunate given that the insights ethnography provides into everyday ways of thinking and acting within organizations  may also help scholars gain a better understanding of how members from different organizations make sense of each other. Whereas ethnography in a traditional sense entails the immersion of a researcher in a particular culturee.g. personally getting to know the employees of an organization in depthmulti-sited ethnography does not confine itself to a single location, but instead follows an object or idea around a multitude of places (Marcus, 1995). Hannerz (2010) provides an overview of how notions of locations have shifted over the years and explains how ethnography has moved from the classical "being there" in a country far away to studying cultures "at home" and, later, to studying "webs of relations between actors, institutions and discourses" (p. 60), which he describes as "studying through". Instead of "being there" the fieldworker is "here and there", studying multiple sites and trying to map the relations, connections, and associations that bind those sites together; this is called a "mobile ethnography" (Marcus, 1995), which traces how culture is constructed across multiple sites rather than within one.
Nonetheless, an important question remains: how should one carry out a multi-sited study? Unlike other types of research, multi-sited ethnography lacks clear methodological guidelines and reflections (Candea, 2007;Nadai and Maeder, 2009;Zilber, 2014). Marcus (1995) proposes various modes of construction to help define a fieldworker's research object and structure fieldwork. These modes of construction are based around strategically selecting a site in order to improve one's understanding of the broader system and around "following" a certain object, idea or process (Marcus, 1995(Marcus, , 2009. In his 1995 article, Marcus proposes six modes of following: (1) the following of peoplethe most common, often seen in migration studiesor of a specific personalso referred to as shadowing (Czarniawska, 2007); (2) the following of an object, such as a report as it makes its way through various departments and organizations (Harper, 1998); (3) the following of the construction and circulation of a metaphor; (4) the following of a story and the way it influences social memory and (5) the following of a biography, which may take a researcher across social contexts. Marcus's last suggestion, which may be most relevant to my research problem, is (6) the following of a conflict, through which the researcher attempts to understand both sides. To analyze the mutual sensemaking processes that, in my study, lay at the heart of the establishment of collaboration, I chose to use and adapt the conflict mode of followingto observe a variety of "conflicting" partners, not only on the "frontstage" where the different organizational actors met but also on the backstage where interorganizational meetings were prepared and evaluated.
In order to follow a variety of partners, one must first gain access to different fields. Multi-sited ethnography poses extra challenges in terms of access and the role of the fieldworker (Marcus, 1995;Nadai and Maeder, 2009;Wittel, 2000). Gaining access to a single ethnographic field is already a tricky process that can be described as a continuous trajectory involving multiple organizational actors and dynamics (Bruni, 2006). A researcher should therefore be mindful of where her research trajectory is taking her and how it intersects "with other trajectories of organizational life" (p. 151), given that all fieldwork experience is partial and dependent on access (Bruni, 2006). In multi-sited ethnography, these concerns multiply as access must often be (re-)negotiated for each field. When it comes to research on interorganizational dynamics, however, and especially when the research field leaves most actors feeling "on guard" as opposed to "at home", this multi-sited character can also provide an advantage. Hannerz (2003, p. 210) describes this situation perfectly when he states that in some sites "there are no real natives [. . .] [t]here are [only] people who, like the anthropologist, are more like strangers." For the multi-sited fieldworker this makes it easier to blend in as both the researcher and the "informants" are strangers to one another.
Following across fields also gives rise to another question: what, exactly, should be considered part of the field(s) and to what extent should a subject be followed across these? Although Marcus (1995) provides constructs for following across multi-sited fields, he does not discuss where or how to "make the cut" (Candea, 2007). The question of what constitutes a field has been debated in anthropology since the last century; for example, in the 1950s when The Manchester School, with anthropologist Max Gluckman as its founder, questioned the ideas behind all-encompassing concepts such as society and culture (Evens and Handelman, 2006).
Gluckman felt that such concepts were too often depicted as something bounded and stable, especially given that, as he argued, every situation is processual and unique (Kapferer, 2006). In his view, researchers are better off using a social situationi.e. an event that unites actors from different backgrounds due to their mutual interestsas a point of entry to its related social situations and the analysis of its underlying social dynamics (1940). Similarly, Gupta and Ferguson (1997) argue against the idea that the world is made up of separate cultures, stating instead that there are no natural points through which one can delimit "here" from "elsewhere" due to the interconnectedness of the modern world. Likewise, they argue that researchers should "decentre" the notion of the field and examine the interlocking of different social-political sites (Gupta and Ferguson, 1997). More recently, Hannerz (2010) reminded us that, in its essence, anthropology is about the relationships between actors, not about the places in and of themselves. In that sense, the field in multi-sited ethnography is more of a political than a geographical location (Wittel, 2000), one that is situational in nature and ties actors from different backgrounds together because of a common problem or interest.
Although a researcher can attempt to clarify who or what should be considered part of the field before starting fieldwork (Zilber, 2014), the boundaries between fields in multi-sited ethnography are inherently fuzzy and can only be further demarcated along the way (Nadai and Maeder, 2005;Zilber, 2014). Not allowing this fuzziness to discourage me, and because my research centred around a much-discussed transformation that needed to be "made sense of" by a variety of parties, I figured that the best way to structure my choices in the field was to follow the collaboration between the medical and the social fields. These parties' sensemaking happened in different ways, causing diverging opinions and conflicting expectations. My main focus, I had decided, would therefore be on (1) municipalities and (2) healthcare insurers, who were in large part the strategic partners responsible for financing the reformed healthcare system. In order to understand how the challenge was understood by the involved parties, I realized that I needed to carry out research in various locations: separately in both parties' "homes" and in the "trading zones" (Kellogg et al., 2006) where both groups met. As I was about to find out, however, this was not nearly as straightforward as I had initially presumed.
In conclusion, even though multi-sited ethnography appeared to be a promising approach to researching the "intersections of organizational peripheries" (Yanow, 2004), it also came with its own challenges. In the remainder of this paper, I will describe my journey across the various fields, discussing how I reacted to three key challenges, i.e. the need to (1) negotiate access, (2) choose between an endless number of potential routes to follow and (3) make sense of a patchwork of data.
Challenge 1. Whose side are you on? Negotiating access in order to follow unfamiliar partners What directors may object to is that they'll have to share their strategy with someone who's also speaking to their negotiation partner. So [observing] the meeting on Wednesday may be even more sensitive than the consultation with the insurer itself. (Email, regional secretary) Although negotiating access is a tricky process in any ethnographic study, it may be even more so in multi-sited ethnography. Due to the segmented nature of the field at large (Hannerz, 2010), access to certain areas does not guarantee access to others. The email citation above shows a key actor's response to my inquiries about joining the meeting of four representative aldermen as they prepared for their first strategic meeting with insurers. Given my interest in the relationship between these partners, and how they would approach collaboration where there used to be none, I felt it was an event I could not miss. The aldermen, however, were hesitant to provide access and I could see why. Because I was taking a multi-sited approach I would not only be conducting fieldwork with the aldermen, but also with their unfamiliar partners: the insurers. I had yet to develop strong relationships in the field. Why should the aldermen trust a random student, whom they did not (yet) know, enough to let her in on their strategy for their first meeting with insurers, with whom their relationship was already troublesome? Especially at the start of my research, I had to continually negotiate and justify my presence at "insider" meetings. In cases such as the one mentioned above, doing so proved particularly difficult. After emailing back and forth the regional secretary concluded that I could come to their office but it (a five-six hour train ride) might be for nothing. Before the start of the meeting I built a case as to why my presence was important to both my research and the actors involvedand, luckily, I was allowed to observe. In other cases, however, my negotiations led to a dead end. On the insurer's side, for example, joining intraorganizational meetings proved impossible. In order to grasp their point of view I was forced to rely on my observations from interorganizational meetings and in-depth interviewsa dynamic that is frequently seen in multi-sited ethnographies (Hannerz, 2003;Nadai and Maeder, 2009).
Gaining access to the various fields was a tricky process. Again and again I was forced to (re-)negotiate my presence at meetings with actors from multiple fields, constantly trying to align their interests with mine. Conscious consideration of what I was allowed to access and why proved critical. At times, actors tried to take advantage of my position in the field. In an attempt to get "in" with a healthcare provider, for instance, one policymaker offered to give me the contact information of a number of district nurses if I would, in turn, tell her what the nurses were doing "right" and "wrong" (which I declined, emphasizing instead the relevance of my research to the field). Another time, my suspicion was triggered by the eagerness with which an actor suggested I visit a collaborative project and indeed: it was clearly an outlier and did not represent the way the project was running in most locations. In fact, one of the organization's local employees with whom I discussed my visit exclaimed, "Oh yes, that's an extreme example she likes to send people to see!" In this case, "people" referred to the financiers, ministry policy advisers and now me. In the end, I gained a much better sense of the organization's everyday life when I followed actors around the organization's other projects.
The problem of actors trying to leverage my cross-field position alludes to an issue that emerged when I was in the field: what was my identity? As I stated earlier, multi-sited ethnography does not allow for any options besides the role of researcher. An added complication in my case was that the interorganizational meetings I attended often included at least some actors I had never met before. Not only did our unfamiliarity require me to constantly re-negotiate my "identity", I could also not assume the obviousness of my position to the different actors involved. Not knowing who I was, these actors often wanted to "place" me and felt the need to make sense of my presence. The times I arrived chatting with aldermen (as valuable as that was) I noticed that the other actors often linked me to municipalitieseven if I introduced myself as being from university later. I felt that this influenced what actors from other organizations chose to confide in me. For example, once I arrived before municipal actors to a meeting focussed on the integration of the social and medical domains, which up until that point had only included actors from the medical field. The consultant facilitating this project, who had a strong opinion about municipalities, told me, "You should never include the [local] government, that's disastrous for integration." Later, when he saw that I was very familiar with the municipal actors who arrived (one of whom would often drop me off at the train station, which meant I stuck around her after the meeting), he tried to soften what he had said earlier. As such (unintentional) cues clearly influenced the data I collected (Ybema et al., 2019), it was important for me to be aware of my positioning in the field, and of how the actors I spoke to made sense of my background and relationships within the field. In light of this, I made it a point to arrive to meetings early so that the actors I had never met before got to know me as an academic researcher rather than as someone associated with a municipality. Also, I made sure to not always sit next to the aldermen (although when I did they often whispered their private thoughts to me throughout the meeting). Not being seen as directly linked with the municipalities enabled me to get a better view of how the other actors made sense of the municipalities with whom they were tasked to work.
Nonetheless, I did become more a part of the municipal field than of the othersthe municipal field was my starting and base point. After about a year of attending strategic municipal meetings my presence was taken for granted. I was both on the invitation list and expected to be there "to strengthen the connection between research and practice". When I encountered aldermen secretly talking in the hallways they would simply continue discussing their true intentions (Was it really about information exchange or mostly about financial investments in the social domain?) for both their collaborations with the insurer and their persuasion tactics; not only despite my presence, they would actually include me in their backstage conversations. To give another example, once, right before a board meeting was about to start, it became obvious that someone had forgotten to print a name tag for me. I did not mind and said I would stay a mystery guest until it was my turn to introduce myself (these meetings tended to include a round of introductions to familiarize all of the individuals involved). One of the aldermen quickly interjected and asked an employee from the regional office to make a name tag for me saying, "Now it looks like she does not belong here and that's obviously not the case." Although I sometimes found it difficult to have to (continually) negotiate and justify my place at the table, I also liked these experiences. Being at the "municipal table" started to feel like a home away from home for me and, later, as I sat in the train heading back to my actual home and reflected on my experiences in the field, I sometimes noticed that I was more susceptible to the municipal side of the story than to the other points of view I had heard. This was often a sign for me to get some distance and reflect on what I had experienced in the field, juxtaposing the stories I had heard at the municipal table with those I had heard from healthcare providers and insurers.
Furthermore, my following of unfamiliar partners as part of my multi-sited ethnography posed additional challenges in terms of access and positionality: how would I gain access to and relate to actors throughout the different organizational fields, and how would these relations affect the stories they shared with me? Such questions of access and positionality forced me to be more reflective about my relationships in the field and the implications these relationships had on the way I made sense of my findings. I made it a point to juxtapose the stories I had heard from different actors and to reflect on how I felt about these stories and why. In this way, a multi-sited approach helped me gain analytical distance from the actors I was studying. Although I had anticipated encountering some access-related challenges due to the segmented nature of multi-sited fields, I did not foresee what proved to be one of my biggest access challenges of all, namely: gaining access to more fields than I could handle.
Challenge 2. Being everywhere and nowhere at once: following a boundless subject After I was finally able to gain access to multiple organizational worlds, the lack of boundaries in multi-sited ethnography quickly became overwhelming. Before long, I was drowning in a sea of possible connections. I felt I had to be everywhere and nowhere at once, not knowing where to draw the line in what to include or when to leave my fields. My supervisors had advised me not to stay in the field for too long, to step back and gain some distance before returning once again. They themselves have even written about "how to resurface" and find a balance between immersion and distance in fieldwork ). I, however, felt this was simply impossible. The fields were too complex and dynamic for me to be able to distance myself and still understand the field upon my return. Change was seemingly constant. Every time I returned to the field the world seemed to have shifted. New local initiatives continuously emerged, turning the field into "a spaghetti of projects", as one alderman put it. Local actors often talked about continuous chaos and change, about desperately needing a dot on the horizon to work towards (or at least a "big stripe"), and about so-called "white areas", which were field partners that were important but still unknown to local actors. Similar to the actors I was following, I also felt hindered by the abundance of chaos, feeling powerless to draw boundaries between what was and was not important for me to follow in my fieldwork. My subjects were boundless.
I was intrigued by how local actors went about building collaborative relationships with unfamiliar partners from scratch as well as by how local actors interpreted their rapidly changing environments. As I wanted to follow regional strategic collaboration between the social and medical domains, my intent was to join every meeting related to this matter and explore how the actors made sense of not only the field, but also themselves, each other, and the changing relationships between them. Before long, however, the question of where to draw the line in terms of what to follow became problematic. It turned out that, despite being strategic financial partners, the healthcare insurers and the municipalities were not actually speaking to each other. Consequently, there was no "natural" field to use as a starting point. Additionally, it quickly became apparent that the mutual sensemaking processes I was looking to study occurred not only during meetings between actors from a single organization (which I had already anticipated joining) but also during meetings with actors from organizations outside of the strategic collaboration in question. Also, as I was busy trying to grasp the context of this collaborationitself the result of a major transformation in the Dutch healthcare system -I knew I could not lose sight of how this change (and the role of the implicated financial partners) played out on a more practical level. This again meant the absence of a "natural" place to draw the line in order to understand the experiences of the collaborating actors.
I was not alone in my difficulty making sense of all that happened in the field. Local actors also struggled and I often heard frustrations about continuously changing ground rules and partnerships. As actors went about trying to identify which organizations they should connect with in order to achieve their goal of seamless social and medical care, relevant and to them previously unknown parties often popped up. As one general practitioner (GP) mentioned: I once joined a project [focused on the transition], I figured it would be nice to know how it worked in that social domain. And it astounded me how many of the people that were there I'd never met before. Those were all people who, in whatever capacity, are working with my patients. (Notes, general practitioner) The healthcare reform was like a meteor that hit the groundit swiftly rearranged the entire healthcare system and local actors were forced to reinvent ways to connect to each other and of deciding which partner was relevant. The quotation above underlines this. According to that GP, there was a whole world out there of which he had been previously unaware. In addition, change was ever-present. For municipalities this meant that the emphasis on collaboration at times shifted from collaboration with insurers to collaboration with other field partners, such as GPs and other healthcare providers. My focus in the field broadened accordingly: I felt that, in order to understand how local actors made sense of the systemic and relational changes in the field, I had to not only trace the connections between the financiers of healthcare in both the social and medical domains, I also needed to look at their connections with the providerson a strategic as well as on an operational level. Concretely, this meant I was shadowing and observing more and more meetings, between more and more partners, on more and more levels, trying to make sense of it all.
Although I stuck with my initial subjectsthe municipalities and the insurers -I was continuously expanding my focus and, as a result, taking on more than I could handle. I travelled across the region to join meetings, talk to policymakers, and shadow nurses during home visits, trying to follow and include everything that seemed like an implication of such a major change in local governance. Eventually, however, I felt I could not keep up with all the places I needed to be andunsurprisinglyended up feeling like I was both everywhere and nowhere at once. It was simply impossible for me to follow all of the collaborations that resulted from the healthcare reform. The size and pace of the field made the metaphor of following, which is meant to structure multi-sited fieldwork, a problem in my researchwhat were the boundaries of my field? I struggled with how to make an informed decision about where to draw the line and worried about missing important events or leaving my field unfinished.
As I tried to grasp the sensemaking processes I observed across the healthcare system, I no longer felt I had any control over the trails I was following. On the one hand I felt as if I was on a train that had gotten stuck in the middle of nowhere while on the other hand I felt as if I were on a rollercoaster, trying to be everywhere at once. My attempts to follow integration in different localities, within different organisations, and on different levels, in an effort to see how integration was constructed across sites rather than within one, left me both confused and travel-worn. These feelings compounded every time I went back into the field until, eventually, I felt unable to think or focus. I was so stuck that the only way forward was to stop my fieldwork entirely. In a sense, being travel-worn from travelling across so many sites had the effect of helping me gain the distance I needed andlacking a natural point at which to stop my fieldworkthis enabled me to then cut off research opportunities and "unfollow" the actors and fields altogether. I cancelled my future appointments and meetings, and went back to my desk at universitytrying to make sense of the patchwork of data I had collected in the three years prior.
Challenge 3. A patchwork of data: making sense of my "unfollowed" fields Reality in the raw can be a pretty formless and meaningless thing: there will always be the need for an interlocutor, someone prepared to take on the job of constructing the rough assemblage [. . .], the process whereby puzzling events are woven into a broader fabric that makes sense of them as some kind of whole. (Bate, 1997(Bate, , p. 1168 There I was, back at my desk with piles of data. The field was puzzling even for the actors in the field and it was up to me to put the different pieces together, and to see if I could construct "some kind of whole", as Bate describes. Although this is challenging in ethnography in general, it may be even more so in multi-sited ethnography given that "multi-sitedness actually means not just sites, but spatialized (cultural) difference" (Falzon, 2009, p. 13). The actors' various versions of reality and the way each made sense of certain events, rules and regulations made it difficult to construct a coherent analysis of how the actors made sense of each other within their rapidly changing environment. Although it was all a bit disheartening at first, I soon remembered why I had chosen to pursue multi-sited ethnography to begin with: to understand how actors across the various fields made sense of their collaboration in the context of the newly enacted healthcare reform. I started analyzing my data in search of things that could help me understand how the actors made sense of both themselves and each other while collaborating across organizational and sectoral boundaries. By coupling this empirical question to theoretical concepts such as boundary work I was better able to structure my data.
During my analysis, the advantages and value of carrying out a multi-sited ethnography also became more and more apparent. Not only had I been able to experience some of the complexity and dynamics that the actors themselves had experienced but having conducted fieldwork in different locales also helped me understand how and when various actors did or did not connect. By reading through my field notes, transcripts and the documents I had collected across the various sites, I was able to map the emergence and evolution of the strategic partners' collaboration. At the onset of my research there had been no history of collaboration between financial partners [1]. As a result, rather than simply joining the interorganizational meetings as I had envisioned, I ended up also following the different fields individually. Doing so, it became quickly apparent that the various actors each had a different understanding of what the required collaboration would entail. The aldermenwho had just become responsible for a substantial part of the healthcare system, including tasks with which they had no prior experience and desperately wanted helpfelt it was important to have regular face-to-face meetings with the insurers in order to share information and discuss joint issues. The insurers, on the other hand, felt the new tasks were the responsibility of the aldermen and did not concern them. In their eyes, there were few tasks that necessitated both parties' teamwork and in most cases virtual collaboration would more than suffice. In other words, it quickly became apparent that both sides of the table made sense of the interorganizational financial collaboration demanded by the changes in the field differently. This mismatch caused a lot of bad feelings amongst actors in the field. Insurers felt unfairly judged and municipal actors felt they had no influence, as if they were talking to a brick wall, forced to surrender to the likes of the insurer. Given that this mismatch could only be uncovered by studying multiple fields rather than one, a multi-sited approach in this case clearly had its merits.
Another aspect of the relationship between insurers and municipal actors which could only be revealed through the study of multiple fields was that the actors both blamed each other for the same things: the other was too inflexible and set in their own ways. Aldermen, for instance, expressed feeling like they were "completely dependent on the insurer" and "did not have a say, and could only wait and see what the insurer came up with". However, this also happened the other way around. For their part, insurers told me that they had tried to set up initiatives but that these had failed because, according to the insurers, "municipalities have very specific policies they want to follow" and "different ideas about what should happen [in the healthcare system]". By this point, it had become clear to me that both parties were mostly focussed on their own visions and interests. Given a lack of dialogue between the various partners, each side ended up constructing their own interpretations of the reformed system, including the required financial collaboration, which in turn led to misalignment and friction between the two partners. Multi-sited ethnography helped map these different constructions and the resulting misalignment and, when I later shared these findings with the actors in the field, they confirmed its pertinence. As one alderman put it, "We all think we're the spider in the web. I'm that spider, but you are too, and you ... But that's impossible, of course." Even though the fields as a whole were too big to grasp and make sense of, multi-sited ethnography helped me map the social constructions I observed in certain elements of the reformed healthcare system on a micro-level. Studying multiple fields enabled me to shed light on how the collaboration between financial partners was built from scratch and why it got off to such a difficult start. Additionally, by juxtaposing the stories from actors from different organizational worlds, I was also able to shed light on how a mismatch in terms of the content partners discussed later led to problems in the practical organization of seamless healthcare. For instance, an often-ambiguous point of discussion concerned each partners' sense of ownership: what belonged to the medical field of the insurer vs the social field of the municipality? The boundaries were blurry and unclear, which led to gaps in the organization of integral social and medical care. It became apparent that, e.g. a medical problem could mean something different to a municipal actor ("someone who is ill") than to an insurer ("someone who requires medication") and the same held true for social problems. Such misaligned constructions provoked extensive discussion and made it hard for local actors to know who was responsible for what.
Although I had initially been puzzled by the patchwork of data I had collected throughout my fieldwork, I eventually owed my ability to shed light on these different constructions and on how they did or did not add up to a whole across the system directly to my adoption of a multi-sited approach. Despite its challenges, a multi-sited ethnography proved both a worthwhile endeavour and the most fitting way to approach the problems central to my research. It forced me to account for the complexity I encountered in the fieldcomplexity the actors themselves had faced on a daily basis, complexity I had now experienced myself. There were no natural boundaries to be found in the field, only the boundaries we drew ourselves (both as researchers and as field actors). As Bate stated back in 1997: "[I]t is time for anthropologists to stop seeing wholes that are not there" (p. 1157). I will elaborate on this argument in the following discussion, where I also share some of the lessons I learned throughout the course of my fieldwork.
Discussion: what did I learn as I followed collaboration across healthcare fields? This paper illustrates my journey across (parts of) the Dutch healthcare system. I have already discussed the three main challenges I encountered during my multi-sited fieldwork and shared what I learned by engaging with these challenges: (1) how the continuous need to negotiate access stimulates reflection on positionality within the field; (2) how the endless number of trails to follow forced me toeventually -"unfollow" my fields and (3) how the resulting patchwork of data revealed that there was no whole to be found or followed, only a plethora of perspectives across sites, which may or may not align. The practice of following is not as straightforward as it may appear. In this discussion, I will further develop my critique of the metaphor of following to further inform future following-based, multi-sited research endeavours.
Although the metaphor of following provides a visual and concrete idea of how multi-sited fieldwork may look, I fear it also makes it appear more straightforward than it is in practice. What troubles me about the metaphor of following is that, much like the metaphor of a train, it makes it seem as if there is a single object to follow. You simply get on the train, get off at the relevant sites, and get back on the train to continue on the tracks. This disregards the fact that "life" goes on at every site even after the researcher has leftboth within the home organizations themselves and during their meetings with other partners. The tracks, then, are not a site that can be holistically described in their entirety, but rather they consist of countless possible and followable trails (see Figure 1). Marcus (1995, p. 102) states that multi-sited ethnography is indeed about "an emergent object of study whose contours, sites, and relationships are not known beforehand", which implies that the object of study will be known afterwards. However, my findings support that the holistic set of contours, sites and relationships of the objects studied cannot be known at all. On the contrary, they remain emergent. In my case, integration between the social and medical domains was a widely dispersed topic that involved many partners. There were no "emerging" boundaries prescribing what and what not to include. In fact, while I was in the field(s), the boundaries only seemed to get blurrier as an increasing number of partners seemed to be related to the issue I was following.
How, then, does a multi-sited fieldworker follow an object with such unclear boundaries? How should one decide which trail(s) to follow? A central theme in these questions is how to delimit the boundaries around the multi-sited field(s): what should I consider part of my research and where should I draw the line? The ongoing horizontal expansion of my fields allowed me less time to stay in one place, meaning less time to dive in vertically and explore a single site in-depth (see Falzon, 2009). I could not keep up with all the places I felt I should be and was stretching myself very thin. Add to that the notion that there is never an ethnographic field out there "awaiting discovery" to begin with (Amit, 2000) and the question of what constitutes a field became a pressing one indeed. Ultimately, the fieldworker is responsible for bounding his or her field and making the cut (Candea, 2007), however much the actors in the field may guide those decisions. This challenge multiplies in the case of following a multi-sited object. Here, there is not a single boundary to be drawn, but rather multiple boundaries across fieldscausing the researcher to constantly weigh the value of spending (more) time on a particular side vs broadening his or her horizon across the organizational fields. Although intended to structure multi-sited fieldwork, this feature of the metaphor of following was a problem for meto what extent should I follow the developments? Postponing drawing a boundary around the field(s) to be included in my research led me to run into my own.
The question of how to draw boundaries around fields essentially brings us back to the issue of what constitutes a field, which, as I discussed earlier, has been a topic of debate for decades (Gluckman, 1940;Gupta and Ferguson, 1997). Given the nature of contemporary societal issuesincluding their increasingly frequent crossing of organizational boundaries, which requires research across fields rather than within onethe need to define what constitutes a field has become an even more pressing matter. Although not necessarily adding up to a clear and bounded definition, scholars have provided insights that may offer some guidance in approaching multi-sited research endeavours. Looking back on the three challenges I myself encountered as I followed collaboration across the healthcare fields, and taking insights from anthropology and organizational ethnography into consideration, I have discerned three corresponding suggestions for future research. Knowing that these lessons are overly practical and have been taught in some form for ages, and without having any illusion that these lessons will eliminate the challenges inherent to multi-sited ethnography, I believe that my findings may help researchers regain some direction when following becomes overwhelming in the midst of multi-sited ethnography.

Challenges of multi-sited ethnography
Carefully delimit the fields to follow. Shortage of access is often considered a problem in ethnographic research. In the case of following multi-sited fields, however, an overload of access may turn out to be the problem instead. Building on Bruni's (2006) notion of access as a trajectory, access trajectories in the context of multi-sited research are also multiplied and, importantly, not always worth pursuing (if not entirely impossible). Moreover, although there is no way to set definite boundaries around events in advance, it may be helpful to start with a socio-political event (Gupta and Ferguson, 1997;Wittel, 2000) or social situation (Gluckman, 1940) and, from there, derive who or what should and should not be considered central to one's research ambitions (see Zilber, 2014). In other words, it may be helpful to consider what you will accept access to beforehand or you may end up "juggling" (Hannerz, 2010) more fields than you can handlewhich, from experience, I recommend trying to avoid.
"Unfollow" the field(s) with an attitude of reflexivity. As I alluded to in my previous point, the field(s) may include many potentially intriguing trails to follow. This can make a fieldworker feel as if he or she "has" to be everywhere and, equally, as if missing something would be tantamount to failing. Realistically, however, a fieldworker will never be able to follow everything, because of competing priorities (Hannerz, 2010) but also because two events may take place simultaneously (Candea, 2007). Accordingly, researchers should release themselves of the need to "be everywhere at once". Likewise, however, this also implies that fieldworkers are themselves responsible for the boundaries they draw in and across the field(s) (Candea, 2007). Although no one can be everywhere at once, we can be transparent and reflective about where we are and why. It may also be useful to do fieldwork in cycles and to (quickly) unfollow one's field(s). Doing so may prevent the researcher from becoming so immersed in a field that she acquires the same blind spots as the native actors she is trying to follow . Stepping back from the field, to study our field notes with some distance, can also enable us to then come back refreshed and sharp. This can be particularly useful in multi-sited ethnography as it can help us decide which field(s) to explore in more depth when different fields appear to be competing for attention. At last, even if "unfollowing" feels like abandoning your research, remember that most ethnographies are likely to be unfinished as life in the fields continues to go on.
Sensibly follow guiding questions and concepts. In his 1995 article, Marcus describes following purely in terms of fieldwork. However, using following as an approach to multi-sited fieldwork carries additional implications for head and text work (Van Maanen, 1998). Given the lack of boundaries that may appear during fieldwork, limiting the number of theoretical concepts may help researchers to both set boundaries and make sense of a patchwork of data. In my case, my initially broad research question and number of theoretical concepts made me resemble the very hungry caterpillar that never has enough: I needed to explore all possible trails as so much seemed relevant to my research question. The consequences of this, however, were that I was less able to explore these issues in depth and I ended up feeling overwhelmed. Limiting the number of theoretical concepts and abductively analyzing my fieldnotes accordingly may have brought more focus to my writing as well as to my observations and decisions around what to follow in the field. This paper has revealed a number of benefits and detriments of the "following" approach that have not been articulated before. First, the need to negotiate access across multiple fields helps researchers become acutely reflective about their varying positions in the field(s). Second, the vastness of the fields (and the accompanying bewilderment) can provide an incentive to eventually "unfollow" one's research objects and actors, and to regain critical distance. Lastly, the process of searching for a holistic account in a patchwork of data can help scholars realize that there is, in fact, no "whole" to be foundwhich, in my case, allowed me to experience firsthand the everyday complexities, irregularities and uncertainties the research participants themselves faced precisely because they, too, were required to act in multiple settings. In these ways, multi-sited