Situating partnership activity, an activity theory inspired analysis of school to school inquiry networks

The issue of how schools work in partnership with each other andwith other organisations has always been of interest to school leaders, policymakers, and practitioners. This article adds to the body of research on partnerships in part by focussing on school to school partnerships and in part through the use of activity theory as an analytical tool. This is based on a study which examined how school partnerships developed, and of what helped and hindered this process. Drawing from activity theory this article presents a “situated” concept of school partnerships emphasising “nested” (within the school) and “overlapping” (between school) features. The successful operation of any partnership, it is argued, requires consideration of both of these components of partnership, and of how they relate to one another. Subjects: Education; School Leadership, Management & Administration; Schools & Schooling

Understanding how schools can work in partnership with each other, and with other organisations, is of widespread interest covering a range of international settings and a variety of types of partnerships (Edge & Khamsi, 2012;Fagan, Brooke-Weiss, Cady, & Hawkins, 2009;Forlin & Rose, Andrew Townsend ABOUT THE AUTHOR Andrew Townsend is Associate Professor in Educational Leadership in the School of Education at the University of Nottingham. He is a member of the Centre for Researching Educational Leadership and Management (CRELM). He is interested in a range of issues of relevance to educational leadership including, but not limited to: educational change; practitioner and institutional development; educational leadership and educational networks. He is especially interested in the participatory features of these aspects of leadership. His current academic career followed a period of work as an educational consultant supporting, researching and evaluating various educational development initiatives. The interests which have guided his work as a consult and academic are informed by his experience as a teacher of science over a period of nine years.

PUBLIC INTEREST STATEMENT
This article reports on a project which looks at the ways in which staff from different schools can work together for mutual benefit. This is not always as easy as it seems as each school has different priorities and existing ways of working. These are influenced by many factors, including the history of the school, its local context and the make up of its staff. The study which this article reports tried to take account of these factors by adopting a theoretical framework called activity theory. This is used to develop a more sophisticated understanding of how work in school partnerships relates to the work of individual schools. In this article, this then leads to a discussion of how this can be used to inform the development of partnerships. 2010). Much of this partnership literature examines how schools work in partnership with further, or higher, education providers (McLaughlin, Black-Hawkins, Brindley, McIntyre, & Taber, 2006;McLaughlin, Black-Hawkins, McIntyre, & Townsend, 2007) often in the delivery of initial teacher training programmes (Edwards & Mutton, 2007;Mutton & Butcher, 2007). But research has also reported on other forms of partnerships, including partnerships between schools and local communities (Warmington et al., 2004) and between schools and other agencies who work with children (Harris & Allen, 2010). A smaller body of literature concerns schools trying to establish productive working relationships with each other (see for example : Dimmock, 2011;Robinson, 2011;Watson & Drew, 2017). This article contributes to this small and relatively under-examined area of partnership literature. This is a significant gap in the existing literature, as school to school partnerships are becoming increasingly important. This is especially the case in the English education system where successive governments have employed policies embedding competition between schools. It is therefore useful to understand how schools can work together in partnership despite these competitive pressures. Examples of this can be seen in the extension of the existing academies programme, and the introduction of so-called "free schools", where schools receive funds directly from government establishing competition between schools for students and the per capita funding that is associated with student recruitment (Gunter, 2011). This has taken place at the same time as significant cutbacks in the funding of, and subsequent contraction in the size of, the education departments of local authorities (Ball, 2012), and the closure of other education support agencies, such as the British Educational Communications and Technology Agency (BECTA, for more see: Ingleby, 2014) and the Qualifications and Curriculum Development Agency (QCDA for more see: Selwyn, 2011).
What this means is that whilst schools have been given more independence in how they can manage their budgets, the available external support services on which they could previously draw have been much reduced (Avis, 2011). This establishes a context in which schools are increasingly being expected to support each other, despite also being competitors. This marks a shift from direct local government involvement in the running of schools to governance through school networks and alliances (Watson & Drew, 2017). Because of this shifting educational landscape, and the greater emphasis on schools working together, understanding school to school partnerships has taken on an even greater relevance. This study aimed to contribute to this need for a more complex understanding of how school partnerships develop. Although they share a common goal, i.e. the education of children, different schools have different histories and cultures. They can also differ in the ways in which they are organised. This article reports on two cases of school networks based in the UK. Each network was formed of multiple school partnerships. These networks adopted different approaches to establishing these partnerships.
This article also develops the use of third generation Cultural Historical Activity Theory (CHAT) as a theoretical framework for understanding school partnerships (Engeström, 2000). Activity theory provides a means for understanding the socially situated nature of the activity and so therefore of partnerships (Engeström, 2010). There are two main reasons why activity theory is suitable for this analysis. Firstly school partnerships by definition are comprised of relationships between different organisations. Activity theory can help to show how the work involved in establishing and running partnerships are related to the social features of work. The second benefit is that activity systems also emphasise, through the concept of boundaries, the extent or limit of activity. It is this feature of activity theory which has been most used in writing on educational partnerships (Saunders, 2006). Forming partnerships requires working at, or crossing, boundaries and activity theory provides a means for better understanding this intersection between activity systems. Together these are used in this article as a means to situate school partnerships. Much partnership research focusses solely on what happens in the work of the partnerships themselves. But this approach adds to the literature by examining how partnerships relate to the partner organisations independently as well as to their collective work.

Networking and school partnerships
The school partnerships discussed in this paper were established as a means by which schools could work with each other for mutual benefit. This differs from other types of school to school partnerships in which a "successful" school leader and their school help less successful schools to improve (Hill & Matthews, 2008;Robinson, 2011), a relationship which has been criticised as legitimising "new forms of dependency, perpetuating existing imbalanced power relations" (Crossley & Holmes, 2001, p. 400). The imbalance of power in this form of school to school relationship makes it arguable whether they are indeed "partnerships" as the terms implies that the partners work as equals. In contrast, the networks which were the focus of this study were established with the aim that all partners would gain equal benefit. In these partnerships, neither partner was portrayed as the expert, and so these partnerships reflect what is referred to by Watson and Drew (2017) as being a form of interorganisational collaboration (IOC).
This article draws from data collected in two such networks, both of which were established during a period which saw an international interest in networking initiatives (Darling-Hammond, Ancess, & Falk, 1995;Earl, Torrance, & Sutherland, 2006;Rauch & Pfaffenwimmer, 2015). The two networks reported here were organisational development networks (Townsend, 2013a) founded on participatory ideals (Townsend, 2013b) which emphasised: voluntarism of membership; choice over development focus and methods; aspirations to enhance agency; and thus ownership over any resulting change (Day & Townsend, 2009;Lieberman & Grolnick, 1996). These forms of participatory networks are seen as providing opportunities for members to collaborate on developments informed by their professional discretion and autonomy. Sachs (2000: 68) suggests that this allows has the potential to allow teachers to "direct the agenda of teacher professionalism".
These kinds of networks are adaptable entities which do not necessarily involve the kinds of sustained and in-depth relationships which would normally characterise partnership working. But one way to think of them is of comprising of a multitude of school to school partnerships which form, adapt, dissolve and re-form according to the needs and interests of the network members. In this a "network" is an overarching arrangement of smaller more sustained partnerships which, because it is highly adaptable, can respond to rapid social or societal changes in ways which inflexible bureaucratic systems are incapable of doing (Lieberman & Grolnick, 1996).
Despite what are believed to be a wide range of benefits, the organisation and operation of networks can be problematic. Some have raised questions about whether they achieve the extent of complementarity expected of them (Edwards & Mutton, 2007;Furlong, Barton, Miles, Whiting, & Whitty, 2000) and there are also suggestions that networking practices do not always spread through member schools in the ways intended resulting, in contradiction to the participatory aspirations, in the creation of new elites. Finally, there is also evidence that the potential benefits of networks can be undermined by emphasising bureaucratic organisation over relationship building (Edwards & Collison, 1996).
Establishing mutually beneficial school to school partnerships can, therefore, be a challenging endeavour with a host of benefits but also a range of potential pitfalls. The aim of this article is to present a situated concept of partnership which takes into accounts both the individual needs of each partner and the nature of their work together. The argument is that partnerships can not be understood solely by what happens in the collaborative setting of the partnership itself. Neither can it be understood by focussing mainly on the needs and benefits of individual partners. But rather partnership can only be fully understood by taking into account the individual settings of each partner, understanding the nature of their collaborative endeavour and appreciating the resulting links between them. This, it is argued, situates partnerships and emphasises that they occur both in the collaborative endeavour of the shared work, and in the settings of each individual partner. This view is developed from concepts associated with activity theory, which are outlined in further detail in the following section.

Situating and developing partnerships
This idea of situating partnerships draws from three Cultural Historical Activity Theory (CHAT) concepts. CHAT is a theoretical construct which links cognition and activity and which developed from the work of Vygotsky (1978) and Leont'ev (1978) (Engeström, 2000). CHAT is an approach to examining and understanding the complexities of the social world, and in trying to appreciate how these complexities relate to human subjects, their individual and collective motivations and the objects of their activity (Engeström, 2000). The central tenet of activity theory is that human actions can only be properly understood in activity systems. Actions are understood not only as linked to individual goals and aspirations, but also to the social context of the actor including: other people; their respective roles; the rules and conventions which govern activity and influence cognition and of the types of tools (physical and mental) which actors can employ. Part of this entails understanding how the use of objects and tools are related to traditions, rules and conventions which arise over time. Activity theory, therefore, provides a way of conceiving of knowledge, actions and attitudes as located in cultural and historical contexts (Nardi, 1996).
The application of a CHAT is often, although not always, associated with graphical or pictorial representations of "activity systems" an example of which is shown in Figure 1.
The concept of the activity system challenges the idea that the relationships between subjects, objects and outcomes is simple and instead shows how other mediating factors influence this relationship. The relationships between subjects and objects are shown to be influenced by community, rules, a division of labour and instruments, all of which interact and combine to influence the outcomes of any given activity system (Engeström, 2001;Leont'ev, 1978;Vygotsky, 1978). Activity theory provides an understanding of motivation, aspiration and actions, which acknowledges the influence of the social and historical settings of activity (for more on this see: Song & Kim, 2016). It does this by showing how subjects are related to the object of their work, and how the different features of activity systems influence that relationship (Engeström, 2010). Engeström (2001) has identified three generations of activity theory through which the model of activity systems has developed. The first originated with Vygotsky's concept of mediation between object and subject (thus three points on a triangle: subject, object and mediation). In the second generation, this was developed to include more specific reference to the importance of rules, community, division of labour and instruments as forms of mediation, as shown in Figure 1. This provided the basis upon which a third generation of activity theory evolved, which sought to understand how activity systems interact. A version of this model of third generation activity theory is shown in Figure 2.
Third generation activity theory is especially useful as this emphases interactions between the activity systems of different partners. Three particular constructs are used here. These are the concepts of boundaries, transfer and relational agency. All have arisen from studies of intersections between activity systems.

Subject Object
Outcome Figure 1. Activity system.

Boundary
As the name suggests the concept of a boundary refers to the limits of activity. It can be used to describe both physical boundaries, like the outer limits of a building, and virtual boundaries, such as the limits of aspects of practice (Fitzpatrick, 2000) or the limits of individual and group knowledge (Bugnon, Arcidiacono, & Perret-Clermont, 2010). As this study is concerned with the ways in which partners from different institutions share, and collaboratively develop, their practices and knowledge, differing forms of the term "boundary" are relevant here. This refers to both the physical boundaries of partner schools and the extent of the knowledge of partners.
In activity theory boundaries are not necessarily either fixed or impermeable and can be crossed. This can happen when individuals or groups move from one activity system to another, for example, when people leave work and go home, or in partnerships where people from organisations meet to work with each other (Saunders, 2006). The conception of the boundary is of a fluid and dynamic limit to the extent of an activity system which can be crossed by individuals and, through interactions between activity systems, also by knowledge or practices. The questions posed by this for situating partnership work refer to the boundaries of the activity systems of the partners, where they are, how flexible they can be, the extent to which they are permeable and how they can respond to the establishment of a partnerships, i.e. relationships between activity systems.

Transfer and knotworking
Transfer and knotworking both arise from the contact that individuals or groups from differing activity systems have with each other at boundaries. Transfer refers to exchanges of knowledge understanding or practices where activity systems intersect. Transfer is a limited notion of what happens at boundaries as it implies a simple exchange between actors across fixed boundaries. Transferring knowledge or practices may well occur but the goal of partnerships is not only for knowledge or practice to be exchanged between partners, but also for the formation of new, or transformation of existing, knowledge or practices through boundary work (Warmington et al., 2004).
The concept of knots and of knot-working extends the idea of "transfer" as a way of thinking about what happens at the intersection of two or more boundaries (Engeström, 2010;Engeström, Engeström, & Vähäaho, 1999). Knotworking suggests that partners from different activity systems interact at boundaries in ways which are not limited by the rules, regulations or normalised practices of each individual intersecting activity system. This presents both a challenge for maintaining productive meaningful activity but also an opportunity for joint learning and the development of new practices (Tuomi-Gröhn, Engeström, & Young, 2003  knotworking then extends this to ask how partnership relationships are formed, how they function and how this influences both the transfer and production of knowledge and practices.

Relational agency
The use of activity theory to understand partnerships raises questions about what happens at boundaries, and how this can be related to the creative, productive, but fluid interactions associated with networking. The concept of relational agency describes the ways in which relationships can transform knowledge and result in new practices. The consequence of working at the blurred edges of activity systems is that groups of people come together to share their work, learn individually and collectively from interactions and develop new objectives for and approaches to their work and life. It is this which is shown in Figure 2. Each partner works in their own activity system, but by interacting and working together in partnership, a third shared object can arise.
The ultimate aim of partnerships is for the differing expertise of individual partners to be shared and harnessed for the benefit of the group. Benefitting from interactions and relationships, in this manner, is termed "relational agency", or "action with others" (Edwards, 2005). It entails more than a mere exchange or transfer of knowledge and practices, but rather arises from sustained collective work. From this partners can experience transformations in the ways that they perceive the object of their work and the nature of their associated practices (Edwards & Kinti, 2010).
Much of the writing about activity systems and partnership is written from the context of interprofessional learning (Guile, 2011), i.e. partnerships between different kinds of organisations. My contention is that these concepts can also be productively applied to partnerships between schools. While the apparent objects of all schools are the same, in that they are concerned with children's learning, their cultural and historical contexts differ. These differences are perhaps less apparent than in interactions between seemingly different sectors like schools and social services. But school to school partnerships nonetheless provide similar opportunities for learning and for enhancing practices, whilst also offering similar, perhaps even more intractable, challenges to overcome, such as the pressures to compete for students and, as a result, financial resources.

Case studies of practitioner research networks
This article draws from two case studies of school networks (Yin, 2003) which were selected as being illustrative examples of the same phenomena. The networks which participated in this study were "selected" because of their distinctiveness in matching the research criteria (Merriam, 1998;Strake, 1995). They were networks of schools which were established to promote interorganisational collaboration (Watson & Drew, 2017) as a means to achieve the development of members (Townsend, 2013a). Having established that these networks met these criteria for selection, the first stage of sampling was achieved through contacting the network coordinators, in their capacity as "gatekeepers" (Tushman and Katz 1980). Once these coordinators had given their consent for this research to proceed in their networks, network members were then approached to take part in three main forms of data collection which included: (1) Analysis of documents related to the operation of the network. As these were produced for the operation of the network, and not for this research, they were regarded as being primary sources of documentary data (McCulloch & Richardson, 2000); (2) A questionnaire which explored their experience of conducing practitioner research or of being a part of a network in which practitioner research was a core activity.
(3) Observations of network events. These included whole network events where all schools came together, and smaller meetings of particular research groups. These meetings were recorded through field notes (Hartas, 2010).
(4) Semi-structured interviews (Hobson & Townsend, 2010;Kvale, 1996) in which participants provided accounts of their experiences of being a member of the network and of undertaking inquiry in school partnership groups.
The two networks reported here were comprised of 14 (Network One) and 13 (Network two) schools. In total 28 interviews were conducted across the two networks. Each interview lasted at most an hour which represents over 21 hours of recorded data. The interview data principally underpins this discussion as it was in the interviews where the complexity of the formation and operation of partnerships was explored in depth. Participants were first asked to give an account of their decision to join the network and then to discuss, with illustrations, the nature of the partnership work that this membership had entailed. Verbatim transcripts were produced from these interview recordings (Hobson & Townsend, 2010;Kvale, 1996). Interviews were then analysed using NVIVO software to generate themes. This analysis examined how these partnerships had formed, succeeded or failed and what the members of each partnership felt were the challenges and successes of their own partnership. This inductive, thematic (Miles & Huberman1994), approach was then contrasted against activity theory to develop the two narratives of nested and overlapping partnerships which follow.

Situating the activity of school partnerships
The following two sections outlines the experiences of the two case study networks. The ways in which these networks organised school partnerships illustrate differing features of partnership work and the ways in which these links to the contexts of partner schools.

Network One: building a nested research activity
The first of the two networks discussed here comprised 14 schools from the same geographical area in England. These schools were mainly primary schools (with pupils aged between 4 and 11) but also included some schools with older children. The aim of the network was for all schools to establish groups of practitioner researchers termed "School Inquiry Groups" (SIGs). The operation of one such group was described as follows: We meet as a group, at least every half term, [the research group coordinator] chairs the meetings… and then we all have an input and discuss things and get given things to do…. it works really well… The results of [research] have always been fed back to staff, more recently the whole school [staff] has been involved in doing the research as well. In the very beginning, the [inquiry group] did all of the research and… the results were fed back but now it is more of a whole school thing. (Network One inquiry group member) The leaders of this network had established practitioner research as its distinctive identity. To that end, each member school organised a practitioner research group which was almost always entirely comprised entirely of teachers. This was a group who worked together in questioning existing practice, gathering data and planning new interventions from what had been learnt.
The development of a new activity, concerned with undertaking research, within member schools was intended to provide a common basis for crossing the boundaries between partner schools in the network. This provided the purpose for partnerships between schools. To help develop these relationships a senior leader in one of the member schools arranged network wide research events. These included two conferences a year, to which all staff in participating schools were invited, regular meetings of inquiry group coordinators and the writing and sharing of inquiry reports. At each of these meetings, members of school inquiry groups were given the chance to share what they had learnt about their focus for and conduct of research. The consequences of these efforts were described by one participant as follows: If another school has said 'we're focussing on this because we see it as something that is particularly important at the moment, or something particularly concerning', if we know that there are other classes or issues in the school that are similar we can put people in contact and… even if we don't make it part of an inquiry we can still learn from other people's inquiries… there is often far too much competitive attitudes between school really, and not enough working together to solve issues and if they have got similar problems, if something works, let us know. (Network One inquiry group member) The establishment of research as a new activity in the schools who were members of this network, provided a common basis on which to develop their school to school partnerships. But the inquiry groups were not themselves a form of partnership as each inquiry group was limited to an individual school. Instead, partnership work was intended to arise from research within schools. When partnerships did form partner schools had a common, new, activity of research, which was associated with the network and which provided the basis for transfers (boundary crossing) of practices and knowledge. Drawing from Engeström (2001) the activity system analysis of this approach to partnership is shown in Figure 3.
Partnerships were established by introducing practitioner research as a new activity. The participants, or subjects, then had a new object to their work, namely to generate knowledge for practice through research. This was linked to their existing work, in the sense that they remained educators, but the practitioners who became part of the partnership groups formed new communities of practitioner researchers, with new rules and tools both linked to the conduct of research. By taking on this role they also changed the division of labour, establishing themselves as a group responsible for research. This network, however, was not without its problems and there were two particular weaknesses in this attempt to situate partnership activity as nested within schools. The first was a problem associated with the creation of new boundaries within network schools. The establishment of a School Inquiry Groups (SIG) within schools could result in the creation of a new boundary (Fitzpatrick, 2000) between the staff involved in the SIG and those who were not. The division of labour in schools in this nested partnership, resulted in the formation of a new community within those schools, with rules and tools derived from practitioner research, who were separated by this differing activity from other members of their own schools (Bugnon et al., 2010). Thus, whilst the partnerships might have been nested in the member schools, they also potentially resulted in the creation of internal boundaries which limited the spread and benefits of learning associated with those research partnerships (Bugnon et al., 2010).
To avoid this some schools established a rotating membership of inquiry groups. This was intended to ensure that all staff at some point were members of this community of inquirers. Others went further and established inquiry groups comprised of all staff in the school.

Outcome
Research informed practice within school.

Instruments
Tools and language of research.

Division of labour
Establish an Inquiry group within each school with the responsibility of undertaking research

Community
Teaching staff within the same school.

Rules
The establishment of the procedures for undertaking research rooted in the norms of the school.

Subject
Object Undertake research and produce knowledge for practice. I think we learned a lot, joining when we did, from other schools … we set up our research up differently having been to the first few meetings where we talked to other schools about what they were doing and we learned… we made the decision to talk to everybody who wanted to be involved and, if we did it as a whole school, it could be done through staff meeting time rather than other times we were trying not to put more onto people and trying to make them meet additional times and after schools and other meetings and if it was an after school meeting therefore it could be built in. (Network One inquiry group coordinator) In this participant's school no new group was formed to undertake the inquiry. But there were changes, through the adoption of research as a whole school activity, in the instruments and rules which all staff employed to carry out their work (Engeström, 2001). The participant above explains how their school spoke to other schools and then planned their own approach based on their experiences. In this partner schools were learning from each other about how to develop partnerships from within their own schools, in doing so achieving the transfer of knowledge about managing practitioner research across school boundaries (Warmington et al., 2004).
This also raises the second problem associated with this network, namely that whilst the "nested" practitioner research groups achieved some transfer across institutional boundaries through the exchange of information and experience at partnership meetings and events, because partnership work was principally nested within members schools, there was little of the kinds of sustained collaboration associated with relational agency, or indeed with knotworking (Engeström, 2010). Knotworking describes the establishment of fluid and changeable groups which provides the potential for partners to learn from each other. The meetings between different partners in Network One seemed principally to be a chance to share experiences of undertaking practitioner research, not for working together more intensively. As a result, there seemed limited opportunities to develop relational agency (Edwards, 2005). Situating partnership within schools in Network One presented challenges in spreading learning between schools, thus reducing potential benefits. Network Two, in contrast, attempted to situate their partnership work differently and attempted to establish relational agency by design.

Network Two: overlapping inquiry groups
Network Two shared a number of the features of Network One. With 13 schools it was a similar size, the member schools were geographically close and was primarily comprised of schools with primary aged children. This network also adopted practitioner research as its distinctive identity, and appeared, on the surface at least, similar to Network One in that there were a range of network events and that groups of teachers from member schools worked together in groups on small scale research projects.
Where they differ is in the ways in which the practitioner research groups were formed. Whilst Network One nested groups within schools, Network Two brought together practitioners from different schools into cross-school research groups. These groups overlapped school boundaries with the (unstated) intention that knotworking could occur by design and that, from that knotworking, new knowledge and practices would emerge (Engeström, 2010). But whilst Network One was seemingly successful in sustaining "nested" practitioner research groups, albeit with some problems around the creation of new, within school boundaries, the success of the "overlapping" groups in Network Two varied greatly. One participant described the downfall of their group as follows: Well I'm one of the groups that's folded… quite a few [groups] had folded and people had not worked well together. One of our problems at the beginning was we felt we were thrown together as a group we didn't choose each other, we were just spread out as names and put together… [for example] there were two people from one school and… they had an inspection and… stopped coming back…. [Then] there was a meeting… and I was the only one representing our group. One member was on maternity leave, which she couldn't help, another lady was on a one-year contract and when her husband went back to Scotland… she went with him which leaves [me on my own]. (Network Two practitioner research group member) Establishing boundary crossing (Saunders, 2006) groups meant that they existed outside the work of the member schools. This established partnerships where the members from different schools came together to collaborate without practices being dictated by the rules and conventions of their own schools. This was, therefore, an attempt to achieve relational agency (Edwards, 2005) through knotworking (Engeström, 2010). Establishing these practitioner research "knots" between schools meant that they were not based on the practices of member schools in the same way as had occurred in the nested Network One. Some network members struggled to cope with this new research activity in addition to the demands of their existing roles and work.
You know, it was only when we got a letter saying, you're going to meet as a group, that you think, oh right, we'd better do something. You know, if we'd have known, we would have laid out timetable of dates, including the communal things when we all meet as a conference or whatever. And you know, [we were told] you need to meet as a group once this term… It's trying to get everybody's diaries together. It's impossible. (Network Two practitioner research group member) Far from being the intended shared boundary crossing sites of knowledge creation and exchange (Warmington et al., 2004), some of these overlapping groups became a separate, and thus competing, form of activity with the teachers' existing work at their schools. One which was sometimes seen as a distraction from the core business of members. The cross-school nature of the research groups in this network, therefore, provided different challenges to those in the nested school groups discussed in the first case. But the problems of the groups which folded were not universal and some overlapping groups were able to sustain their partnerships and succeed where others had failed.
We were put together as a group. Ours was a big group but nobody dropped out. We saw how there were other nebulous enquiries and decided that we weren't really going to get anywhere with that approach. As a group [we] sat down and talked about what we were going to do, and everyone did actually get on very well together… We went to visit other schools… [and] saw where their… problems were… you can see them, talk about them [and] can discuss how to deal with them. (Network Two practitioner research group member) This group achieved its success by abandoning the topic which they had been given and focussing instead on establishing relationships between the group members. In reference to the analytic activity system (Engeström, 2001), they invested time in forming communities. Having established these relationships, the group then started working on identifying a common agenda, i.e. developing a common understanding of the shared objects of their work. I can't remember what topic people were given but we decided that we [would] meet as a group first and foremost and decide what we wanted to do. We sat as a group and talked about things that we were interested in, in the school and things that we thought children found difficult, things they found easy… We thought transition, within school transition, would be a good thing to look at. We were a real mixture… [with members from] nursery, reception, year 2, year 3 and somebody who didn't have a classroom [but] we wanted something that was going to be used for everybody. (Network Two practitioner research group member) Thus, the object of this overlapping group which had successfully sustained their work was not, at least initially, focussed on addressing the topic they had formed to research. Instead, their first objective was to build relationships and try to understand each other's contexts and interests. It is from this form of relationship building that some evidence of transfer (Warmington et al., 2004) in these overlapping groups became more apparent.
On the enquiry I would say the best for teachers was finding out what goes on in the other schools, how they address problems, they have similar problems but they have a different way of looking at it. That's the thing. It's just a different way of looking at it… People looking at it and saying [to each other] why don't you do this? Then they're thinking oh yeah we could do it exactly like that, but we could try just a bit of that [approach]. It's working with the other teachers, you're really getting to know them and having the freedom to be able to say what you want. (Network Two practitioner research group member) The tensions some of these teachers experienced between their overlapping practitioner research group, and the schools that they worked in, were not therefore insurmountable. This also suggests that boundary crossing (Saunders, 2006) cannot be achieved by putting people from different organisations together, but rather that it emerges from sustained relationship building and mutual understanding. Boundaries define the limits of work of staff involved in the partnerships (Fitzpatrick, 2000). Just putting teachers together to work in partnerships does not instantly give them a new reference point for their work, they remain a part of their existing activity systems. The ideals of knotworking (Engeström, 2001), the fluid creative interaction between partners, is only possible once the members of the new partnership have had time to form a new community. The success of the second overlapping group discussed above, arose because they spent time developing a location for their work which was "nested" in that it took account of their original working contexts and then developed overlapping features to their partnership work. They found a means to achieve what had been achieved by design in Network One, i.e. a direct link between the work of the research groups and their work in schools. The less regulated and potentially dynamic nature of these re-configured overlapping groups can be considered as a form of knotworking (Engeström, 2001). The main aim is not to satisfy a predefined specific focus for development, but to build relationships, and from that to identify common, shared or complementary interests from which relational agency (Edwards, 2005) could arise.

Developing a situated understanding of school partnerships
These two network case studies illustrate different approaches to establishing partnerships. In Network One the partnership was rooted in commonly adopted practices in partner schools, and so was "nested" in the practices of those schools. The adoption of research, a new activity for staff in schools in the network, did not involve partnerships with other schools, but it did provide a common experience on which partnerships could grow. Research provided a distinct identity, this was not something that the schools had done independently and so it became a common experience for schools associated directly with their membership of the network. Partnerships emerged from the nested activity of research.
In the second case, the network took the decision to manufacture cross school partnership groups which were "overlapping", i.e. across institutional boundaries, by design. Network Two also established research as a new activity in member schools. But their approach differed from Network One in that the school to school partnerships were the research groups. In Network One partnerships were the places where within school research groups shared their work with each other.
This raises issues about how school partnerships can be situated, as it seemed that even though some existing boundaries were crossed, and the transfer was occurring across boundaries in partnership groups, the activity which can be the basis of partnership for some, can create additional boundaries which exclude others. This limits the spread of involvement in, influence over the operation of, and benefits arising from partnerships and can provide some explanation for the previous observations of networks whose benefits (and influences) failed to extend to all staff in members' schools (Edwards & Mutton, 2007;Furlong et al., 2000). This concept of nesting shared activity in the work of individual partners also further emphasises that knotworking at boundaries (Engeström, 2010), can not be considered to be independent of the work of each partner and should not also be understood as being limited to the interactions between those partners. Rather they rely, and so are in some respects dependent for their success, on existing work within each partner organisation. A failure to take account of this "nested" feature of partnerships means that the potential benefits are not realised as the partnership might fail to be relevant to members and might be superseded by their institutional pressures.
The nested qualities of partnerships in these two case studies also raises issues about the status of research or inquiry in the practices of teachers. As already discussed, the adoption of practitioner research (McWilliam, 2004) by these networks could be seen as the establishment of a new activity. The activity system shown in Figure 1 has five features (Engeström, 2001). One of these concerns the community involved in any given activity. By establishing research as a new activity in schools, and in cross-school partnership groups, the networks ran the risk of creating new "practitioner researcher" communities. This established new tools, including language, of research as well as a new object, i.e. the production of knowledge through research. But one of the problems with this is that not everyone in a school community was a part of this new activity system. This, therefore, ran the risk of creating new communities, with new boundaries between those involved in research and those who were not. This emphasises a weakness in the applications of practitioner research in these networks in which both networks encountered difficulties in preventing practitioner research becoming an isolated activity. This is one of the critiques of Cochran-Smith and Lytle (2009) and others who suggest that if research is only linked, and not a part of, practice then it will only be something in which a few practitioners are involved. This inevitably means that communities of practitioner researchers become separated from others who are not involved in research (see for example : Lieberman, 2000;Sachs, 2000). The efforts of both networks suffer from this in different ways. But an alternative approach, one suggested by Cochran-Smith and Lytle (2009) entails incorporating research practices into the activity of teaching. This would mean that inquiry was a common part of everyday practice, one which could then provide the nested basis for productive partnerships.
Situating partnership activity in this manner also entails some consideration of the overlapping features of partnerships, the points where partners come together to interact and collaborate. Drawing from these case studies, it seems that knotworking cannot be made entirely by design. It depends on the nested features of the partnership and arises from mutual understanding, the establishment of shared interests and the opportunity to engage in sustained, shared work. Depending on how this is managed the process of working at the intersecting boundaries of two organisations (Bugnon et al., 2010) can provide the basis upon which the fluid, flexible, working partnerships of knotworking can evolve (Engeström, 2010).
The shared nested activity of practitioner research gives a common experience and language upon which boundary crossing relationships can evolve which can, in turn, lead to the transfer of knowledge and actions and ultimately to a basis for relational agency (Edwards, 2005). By presenting partnerships as knotworking across institutional boundaries, the adoption of an activity theory perspective emphasises what these networks had encountered in their work, that partnership is not a simple question of arranging occasions to meet and deciding on a focus for partnership. Instead, it suggests that partnerships entail some emergent features in which the benefits of the creative intersection of activity systems can be realised by building on the partners' respective activity systems. In this respect, the nested and overlapping features of partnership provides both a tension for the development of meaningful partnership activity and the potential for partners to build on their existing work through the relational agency which can be achieved as a result.

Conclusion
The intention of this article is to try to develop, from the experiences of two case studies, and by drawing from third generation activity theory (Engeström, 2001), a way of situating school partnerships. The focus on school partnerships is rare in literature, and the use of activity theory as a way to theorise school to school partnerships is even rarer, and it is this gap in the literature which this article is intended to fill. From this emerged the "nested" and "overlapping" features of, and approaches to, school partnerships outlined in the accounts of the two cases above. Using the concept of activity systems in relation to partnership work (Engeström, 2000) situates partnerships in social and cultural contexts. This emphasises the role and influence of the different aspects of activity systems, i.e. the community which share the activity system, their shared rules and common instruments, the decisions about the division of labour associated with those systems, and the interaction between those activity systems.
This issue of how school to school partnerships are established and operate might seem a marginal issue of relevance only to the committed few who are determined to initiate and sustain partnership working despite the separating pressures of competitive neo-liberal education markets. However, given the reduction in opportunities for inter-organisational support in the English education system which had previously been provided by organisations like local authorities, there is a more significant question, namely how to ensure that beneficial education practices do not remain isolated within particular organisations. There is evidence that educational markets limits schools' opportunities to collaborate and learn together (Jopling & Hadfield, 2017).
Contrary to the suggestions of some, there is also evidence that collaborative relationships can benefit schools (Hargreaves & O'Connor, 2017). In both of these case studies, successful partnership working resulted in beneficial changes in practices in member schools. Not all the attempts at establishing partnerships reported here were successful. Where partnerships failed it was because insufficient attention had been paid to the local contexts of schools. In the partnerships reported here the establishment of collaborative knotworking was conditional on a common purpose established through mutual understanding (Engeström, 2010) What is needed to achieve this is perceiving partnership as both beneficial to, and an extension of the situated practice of, individual schools.
Research is not always believed to be an appropriate activity for teachers (McWilliam, 2004). But it seems, from the evidence of these networks that the development of practitioner research communities provides one potentially useful focus for the development of such partnerships. This can be one means by which teachers can share experiences, share in a process of inquiry and benefit from learning with and from each other (Cochran-Smith & Lytle, 2009). This is, by definition, a form of boundary crossing which can result in relational agency (Edwards, 2005).
Whilst this study has shed some light on some of the complexities of establishing and benefitting from partnerships there is much more than can be done on this topic. School communities are by definition boundaried (Fitzpatrick, 2000). The development of partnerships can provide a means by which schools and educators can cross boundaries and learn from each other (Edge & Khamsi, 2012). But what is needed are more studies which examine partnerships from within organisations. This article has done a little of this, but more sustained research on the effects of partnership within schools, and of how partnerships relate to policy contexts, would add significantly to our understanding of how they can be established and how they can benefit partners and ultimately learners.

Funding
The author received no direct funding for this research.