The Mangle of Practice: Time, Agency, and Science

Andrew Pickering

The University of Chicago Press

Chicago and London


Preface ix

1 The Mangle of Practice / 1

1.1 Science as Practice and Culture / 1

1.2 Representation and Performativity / S

1.3 Agency and Emergence / 9

1.4 The Mangle of Practice / 21

1.5 More on the Mangle / 27


2 Machines: Building the Bubble Chamber / 37

2.1 Building the Bubble Chamber / 38

2.2 The Mangle and Material Agency / 50

2.3 The Mangle and Intent / 54

2.4 The Mangle and the Social / 58

2.5 Actors, Interests, and Constraints / 63

3 Facts: The Hunting of the Quark / 68

3.1 The Hunting of the Quark / 71

3.2 Emergence and Posthumanism in Empirical Practice / 90

3.3 Multiplicity, Heterogeneity, and Association / 93

3.4 Representational Chains / 96

3.5 Discipline / 101

4 Concepts: Constructing Quaternions / 113

4.1 Disciplinary Agency / 114

4.2 From Complex Numbers to Triplets / 121

4.3 Constructing Quaternions / 126

4.4 Concepts and the Mangle / 139

4.5 Science and the Mangle / 144

4.6 Postscript: Mathematics, Metaphysics, and the Social / 147

5 Technology: Numerically Controlled Machine Tools / 157

5.1 Numerically Controlled Machine Tools / 158

5.2 The Mangle and the Social / 165

5.3 The Mangle, Social Theory, and Limits / 169


6 Living in the Material World / 179

6.1 Realism / 180

6.2 Incommensurability / 186

6.3 Knowledge and Us / 192

6.4 Objectivity / 194

6.5 Relativity / 201

6.6 Historicity / 208

7 Through the Mangle / 213

7.1. Antidiscipline A New Synthesis / 214

7.2 Cultural Studies and the Mangle / 217

7.3 Performativity and Historiography: The Big Picture / 229

7.4 Macromangling / 234

7.5 Postscript: Nonstandard Agency / 242

7.6 Postscript: The TOE Mangle / 246

References 253

Index 275


One good thing about writing a book is that you get two tries at the introduction. Chapter 1 is a self-contained preview of the issues that concern me, how I propose to address them, and what I think my examples show. Here I take a more biographical route. It might help in reading the book to have an idea of how it arrived at its present shape.

I have been fascinated for a long time by knowledge. Since my schooldays, I have wondered about how knowledge relates to the world-- about the problematic of realism, as philosophers call it. Having gained quite a bit of knowledge, eventually as an elementary-particle theorist, in 1976 I joined the Science Studies Unit at the University of Edinburgh, where I discovered that there were other people like me. The unit was one of the centers of a small community of people developing a sociology of scientific knowledge, and I tried as best I could to join in the absorbing but very difficult arguments that swirled around that topic. I worked through several case studies of my former discipline and finally wrote a book on it, Constructing Quarks (1984). By the time of that book, I had come to the conclusion that it was no use trying to think about knowledge in isolation. To understand why people believed what they did, it seemed one had to understand how specific items of knowledge fitted in with the practice of their producers and users. Indeed, Constructing Quarks tried to display the historical development of particle physics over a twenty-year period as instantiating a simple model of scientific practice. After the book, I was left with one nagging concern. The problem I started with had not been solved. I now felt clear enough on how knowledge related to people, but not on how it related to the world (except that it was implausible to imagine that scientific


knowledge literally corresponded to the inner constitution of nature). The problematic of realism was still with me.

The important development for me came when I spent 1986-87 as a member of the Institute for Advanced Study in Princeton. Shortly after my arrival, I set out to respond to an essay review of my book by Yves Gingras and S. S. Schweber (1986). They accused me, wrongly, of Duhem-Quine abuse--of mistakenly presupposing a set of arguments associated with the names of those philosophers. Thinking about this accusation led me to an appreciation of a fact that was already emphasized in constructivist sociology of science, namely that scientific culture is not a unitary, monolithic thing (say, a single big theory, as is typically taken for granted in Duhem-Quine-type arguments); that scientific culture is, in fact, an assemblage of multiple and heterogeneous elements. And I realized that I could, on this basis, at last say something about realism. An earlier study I had made of the history of quark-search experiments could be read as an analysis of knowledge production in terms of the difficult and uncertain work of making associations between heterogeneous--in this case, material and conceptual--elements of scientific culture. An essay followed, "Living in the Material World" (1989), in which I extended my previous account of the experimental work of the physicist Giacomo Morpurgo as an exemplification of a "pragmatic realist" perspective on knowledge, a perspective that deserved to be called realist precisely because it specified the nontrivial links that scientists fashion between representations and the world.

"Material World" was important to me for several reasons. First, as just mentioned, it addressed a problematic that had been with me for years. Second, thinking about cultural multiplicity and heterogeneity helped me to see what was original in the actor-network approach to science studies then being developed by Michel Callon, Bruno Latour, and John Law. I have continued to learn from the actor-network ever since. And third, on returning from Princeton to the University of Illinois, I realized that the essay could also be read as significantly extending the analysis of practice developed in Constructing Quarks. In the book, I had argued that scientific practice is centrally a process of modelling, but, looking back, I had failed to appreciate sufficiently the open-endedness of modelling. One can try to extend a given scientific culture in an indefinite number of ways. What the history of Morpurgo's experiments showed was that, to put it crudely, most of those ways do not work. When Morpurgo sought to extend the material and conceptual strata of his culture, the bits did not usually fit together. "Resist-


ances" continually arose in his work relative to the material-conceptual alignments he needed to achieve to produce facts. And, from the indefinite range of possibilities, certain specific modelling vectors were singled out in his practice precisely in that they did issue in such alignments. Practice as modelling, I thus realized, has an important real-time structure, with the contours of cultural extension being determined by the emergence in time of resistances, and by the success or failure of "accommodations" to resistance. This temporal structuring of practice as a dialectic of resistance and accommodation is, in the first instance, what I have come to call the mangle of practice. The mangle and its implications and ramifications in philosophy, social theory, and historiography are what this book is about.

Having arrived at the idea of the mangle, I began to wonder about its generality. The mangle operated, as I then understood it, at a level of detail not usually accessible to empirical study, but eventually I found and started working through the examples that follow here. From this work, it appeared that the mangle could go a long way, and thus, when I set off from Illinois to spend 1992- 93 in the Department of History and Philosophy of Science in Cambridge, my ambition was to write a book about it. Then one last twist entered the story. Morpurgo's quark-search experiments had been the test case for the development of my ideas since I first saw how they could lead into the problematic of realism. For the book, I intended to extend my analysis of those experiments back into their earliest phase, and there I found myself on uncertain ground. I could not persuade myself of the story that I wanted to tell of Morpurgo's early struggles to get his apparatus to work. It was possible to talk about those struggles along lines that were becoming familiar-- in terms of the making and breaking of associations between multiple and heterogeneous cultural elements--but something seemed to be missing. And that something, it appeared, was material agency. In building his apparatus, Morpurgo was trying to get the material world to do something for him, and this needed to be stated out loud. Having recognized that, my project reconstituted itself.

As discussed in the text, talk of material agency has always been suspect in the sociology of scientific knowledge, but not so in the actor-network approach. There, much is made of material agency and, further, of its symmetrical relations with human agency. It was clear, then, that if I wanted to talk about material agency I had better think what I wanted to say about human agency, too. In Cambridge, that led me in all sorts of directions: into considerations of the intentional structure of


human agency, of the scale of and relationships between social actors, of the disciplined nature of scientific work, and so on. At this late stage, I gained a new appreciation of writers as diverse as Michel Foucault and Michael Lynch and a heightened (but also critical) regard for the insights of the actor-network. And the book, in its turn, developed a vital interest in human and nonhuman agency, in how they temporally intertwine, and in how knowledge engages with them. I have thus ended up with a book that I never intended to write, in which an original preoccupation with knowledge has been subsumed into a wider preoccupation with human and nonhuman agency, and in which, as explained at the end, science itself appears as subsumed within a wider field of machinic production and destruction. I am no longer puzzled by how scientific knowledge relates to the world, nor by how it relates to scientific practice; now I feel the need to understand the disciplined, industrialized, and militarized, technoscientific world in which I have lived my life, and how it ever got to be this way.

Thus the history of the mangle. I hope it will help readers come to grips with the form and content of the chapters that follow.

It remains for me to acknowledge my many debts. As mentioned above, this book began in Princeton, at the Institute for Advanced Study. It also ended in Princeton, but at the university. In 1993-94, I completed and made final revisions to the manuscript as a fellow of the Shelby Cullom Davis Center in the Princeton History Department. In between, much of the work was done at the University of Illinois at Urbana-Champaign, and for their friendship, support, and critical acumen I thank my colleagues in the Sociology Department there, my students (especially those who took part in seminars where I stumbled through the earliest versions of arguments presented here), and, above all, the members of the informal seminar of the History, Philosophy, and Sociology of Science graduate program. The HPSS seminar was the visible manifestation of a social and intellectual community without which my work would have been impossible. Also at Urbana, participation in the Unit for Criticism and Interpretive Theory was my introduction to great swathes of contemporary thought, and the unit has to take its share of responsibility for the wilder ideas expressed in what follows. Two periods of support from the History and Philosophy of Science section of the National Science Foundation marked watersheds in my project. A grant in 1989-90 helped me find the time to escape from the detailed analysis of scientific practice and to start thinking about its implications, and a sec-


ond NSF grant enabled me to spend my sabbatical in Cambridge in 1992-93.

Now for the tricky part. More people than I deserve have helped me along the way to this book, reacting to talks, essays, chapters, entire manuscripts, sharing their ideas, yawning, telling me when I was talking rubbish, in correspondence, offices, seminar rooms, restaurants, homes, bars, and pubs. Increasingly over the past few years I have counted myself exceptionally fortunate that my thought and writing has been situated in a very rich and stimulating field of conversations. I therefore have no confidence whatsoever that I can generate a complete list of the individuals to whom I should express my gratitude. But still I should try, and I hope to be forgiven for what will no doubt prove to be appalling omissions. In footnotes to the text, I acknowledge specific debts, but here I offer my thanks in general to Susan Abrams, Brian Baigrie, Davis Baird, Barry Barnes, David Bloor, Geof Bowker, Nancy Cartwright, Soraya de Chadarevian, Harry Collins, Natalie Davis, Norman Denzin, Irving Elichirigoity, Paul Feyerabend, Owen Flanagan, Paul Forman, Peter Galison, Dilip Gaonkar, Gerry Geison, Yves Gingras, Laurel Graham, Ian Hacking, Mary Hesse, Joann Hoy, Piet Hut, Robert Alun Jones, Vera Ketelboeter, Yiannis Koutalos, Martin Krieger, Thomas Kuhn, Bruce Lambert, Michele Lamont, Bruno Latour, John Law, Peter Lipton, Michael Lynch, Michael Mahoney, Peter Miller, Giacomo Morpurgo, Malcolm Nicolson, Ted O'Leary, Ronald Overmann, Trevor Pinch, Michael Power, Diederick Raven, Joseph Rouse, Simon Schaffer, Sam Schweber, Steven Shapin, Otto Sibum, Barbara Herrnstein Smith, Betty Smocovitis, Leigh Star, Adam Stephanides, Fred Suppe, Peter Trower, Stephen Turner, Adrian Wilson, Norton Wise, and Alison Wylie.

From that list, I should single out two people. Since our days together at the Institute for Advanced Study, Barbara Herrnstein Smith has offered me consistently perceptive advice and incisive criticism. The trajectories of our research run along intersecting lines; I have learned much from her writings, and even more in conversations and arguments with her. The other person is Simon Schaffer. At the beginning, it was he who encouraged me to write "Living in the Material World." Almost at the end, Simon, as acting head of department, made possible my sabbatical in Cambridge and, together with Anita, helped make it such a pleasure. While I was there, he somehow found time to read and comment upon not one but two quite different drafts of this book. And more than that, he typically displayed a clearer grasp of where I was going and how to get there than I did. It was an education to have the chance of interacting


with him over an extended period of time. So, thanks, Simon and Barbara. Thinking of Cambridge, I also want to express my gratitude to my old friends Jim and Rhonda and Lee and Paula for their help above and beyond the claims of friendship in the hassles of transplanting me, Jane, and the children back to our native land. I recall remarking on several occasions, and only partly in jest, that they had saved our lives. Most of my writing in 1992-93 was done at Lee and Paula's old dining table, though Paul Waldmann's carpenter's bench deserves a mention, too.

Lastly, as ever I thank Jane F.--oh, for everything.

I think by writing, and my route to this book has been marked by a trail of essays, first stabs at stories and analyses that, for the reasons described above, especially my newfound concern with questions of agency, appear here and there in the text, transformed, redistributed, and accompanied by new material. The essays that have seen print, and the chapters to which they relate, are "Living in the Material World: On Realism and Experimental Practice," in D. Gooding, T. J. Pinch, and S. Schaffer, eds., The Uses of Experiment: Studies of Experimentation in the Natural Sciences (Cambridge: Cambridge University Press, 1989), pp. 275-97 (chap. 3); "Knowledge, Practice, and Mere Construction," Social Studies of Science 20 (1990): 682-729 (chaps. 6, 7); "Objectivity and the Mangle of Practice," Annals of Scholarship 8 (1991): 409-25 (chap. 6); "Constructing Quaternions: On the Analysis of Conceptual Practice," co-authored with Adam Stephanides, in Pickering, ea., Science as Practice and Culture (Chicago: University of Chicago Press, 1992), pp. 139-67 (chap. 4); "Anti-Discipline or Narratives of Illusion," in E. Messer-Davidow, D. Shumway, and D. Sylvan, eds., Knowledges: Historical and Critical Studies in Disciplinarity (Charlottesville: University Press of Virginia, 1993), pp. 103-22 (chap. 7); "The Mangle of Practice: Agency and Emergence in the Sociology of Science," American Journal of Sociology 99 (1993): 559-89 (chaps. 1, 2); and "Beyond Constraint: The Temporality of Practice and the Historicity of Knowledge," in J. Buchwald, ea., Scientific Practice: Theories and Stories of Physics (Chicago: University of Chicago Press, 1995) (chaps. 2, 5, 6).

ONE: The Mangle of Practice

[T]here is at all times enough past for all the different futures in sight, and more besides, to find their reasons in it, and whichever future comes will slide out of that past as easily as the train slides by the switch.

William James, The Meaning of Truth

Desire only exists when assembled or machined. You cannot grasp or conceive of a desire outside a determinate assemblage, on a plane which is not pre-existent but which must itself be constructed . . . In retrospect every assemblage expresses and creates a desire by constructing the plane which makes it possible and, by making it possible, brings it about . . . [Desire] is constructivist, not at all spontaneist.

Gilles Deleuze and Christine Parnet, Dialogues

This is a book about science that ventures into the worlds of mathematics, technology, and the workplace. It offers a general analysis of scientific practice, which I call the mangle, and some pointers as to how it might be extended toward an understanding of the reciprocal production of science, technology, and society (STS).l It is also a book about time and agency that addresses central questions in the philosophy, social theory, and historiography of science and beyond. This chapter lays out some basic features of my position; the rest of the book consists of examples and articulations.


Science studies has been an exciting field over the past few decades, and one source of this excitement has been a continual expansion of conceptions of science as an object of study.2 Until the late 1950s, it seemed


1. I hope, therefore, that "science" will be read hereafter as an umbrella term of a greater than usual extent.

2. The core fields of "science studies" are history, philosophy, and sociology of science but, as indicated in the previous note, I construe the term broadly to encompass both


enough to think of science as a body of knowledge, a collection of empirical and theoretical propositions about the world. This body of knowledge constituted a self-contained topic for the philosophy of science, for example, whose job it was to enquire into, and possibly to legislate upon, the formal relations between the propositions it contained. But the work of Norwood Russell Hanson (1958), Thomas Kuhn (1970) and Paul Feyerabend (1975) changed all that.3 Especially Kuhn's persuasive periodization of the history of science into stretches of "normal science" separated by "revolutionary" gulfs challenged the self-containment of the science object, and opened the way for new waves of scholarship to wash over and reconceive it. Thus, since the 1970s, work on the sociology of scientific knowledge (SSK) has increasingly documented the importance of the human and the social in the production and use of scientific knowledge.4 Social structure, social interests, human skills--all of these have come to be seen as constitutive of science, as integral to science in interesting and important ways. Further, though SSK's primary focus has been, as its name states, on knowledge and the social, empirical work in SSK has served also to foreground the material dimension of modern science--the omnipresence of machines, instruments, and experimental setups in scientific research. This dimension had long been ignored in mainstream history and philosophy of science, but here SSK has made contact with an alternative philosophical perspective powerfully articulated by Ian Hacking (1983), which seeks precisely to emphasize the machinic aspects of science. Finally, from the late 1970s to the present, scholars have evinced an increasing interest in the details of the day-to-day doing of science. This interest has served further to expand our conception of the science object by documenting its sheer multiplicity and heterogeneity. All of the dimensions of science


contributions from other disciplines interested in science (anthropology, political science) and work that extends into the study of mathematics and STS.

3. Suppe (1977) surveys developments in philosophy of science consequent upon the work of these authors.

4. Canonical books in SSK include Barnes 1974, 1977, 1982; Bloor 1991, 1983, Collins 1992; and MacKenzie 1981b. See also Shapin (1982) for an excellent review of the history and sociology of science literature from an SSK perspective. I should explain why I single out SSK here, rather than the preexisting Mertonian approach to the sociology of science (for a comparative review of both, see Zuckerman 1988). Although the latter did expand our conception of science as an object of study by exploring its institutional structure, it did not envisage the detailed intertwining of the social with the technical in science that is a major concern in both SSK and this book. The Mertonian approach, as it is said, has been more a sociology of scientists than of science.


just mentioned--the conceptual, the social, the material--have to be seen as fragmented, disunified, scrappy.5

I can now sketch out the problematic of this book by, first, reexpressing what has just been said in terms of an expansion of our concept of scientific culture. Whereas one could once get away with thinking of scientific culture as simply a field of knowledge, in what follows I take "culture" in a broad sense, to denote the "made things" of science, in - which I include skills and social relations, machines and instruments, as well as scientific facts and theories. And then I can state that my abiding concern is with scientific practice, understood as the work of cultural extension. My problematic thus includes the traditional one of understanding how new knowledge is produced in science, but goes beyond it in its interest in the transformation of the material and social dimen-sions ot science, too.

Two points of clarification should immediately be entered. One is that in this book I seek a real- time understanding of practice. I want to understand the work of cultural extension in science as it happens in time. This is to be contrasted with retrospective approaches that look backward from some terminus of cultural extension and explain practice in terms of the substance of that terminus. The exemplary instance of the latter is what I call "the scientist's account" (Pickering 1984b), in which accepted scientific knowledge functions as an interpretive yardstick in reconstructing the history of its own production. I think that there are serious historiographic problems in such retrospective accounting for science (Pickering 1989b), but rather than rehearse them here, let me just note that my project is a different one and that, for my purposes, to indulge in retrospection would be circularly self-defeating and must be eschewed.

My second point of clarification takes us back to the recent history of science studies. It is probably true to say that many authors engaged in exploring the work of science once shared my interest in time--back


5. Several streams of work helped to constitute the doing of science as a topic for research in its own right. Within traditional history of science, Holmes (1974, 1981 1985) has pursued the theme most tenaciously (an important recent study is Kohler 1994), but perhaps the most influential source has been the development of ethnographic studies of "laboratory life": Latour and Woolgar 1986; Knorr-Cetina 1981; Lynch 1985a. From another angle, the interest in scientific work appears as a continuation of the concerns of SSK: Pickering 1984b; Gooding 1990. Yet another source has been pragmatist studies of science (which take the work of Howard Becker and, especially, Anselm Strauss as their point of departure): see Star 1991b, 1992; and Fujimura 1992. For surveys of current perspectives, see Pickering 1992b; and Clarke and Fujimura 1992.


in the early 1980s, say. But as I have been writing this book, it has dawned on me that a kind of purlfication has taken place. Much of the most interesting work now being done is not concerned with practice as I have just defined it, but takes the form of atemporal cultural mappings and theoretical reflections thereon. My present interest in the temporality of cultural extension leaves me, I think, in a minority as far as current initiatives in science studies are concerned.6 I say this not in a spirit of critique of what I call the cultural-studies approach--in fact, I draw extensively upon its findings as the book goes on, and I argue in section 7.2 that it is complementary to my own approach--but to make clear the tendency of this book. And in this connection, I think it will help at this stage to note an ambiguity in the word "practice," an ambiguity that explains why it took me, at least, so long to realize that my project had diverged from others.

One sense of "practice" is the generic one around which all that follows is organized--practice as the work of cultural extension and transformation in time. The other sense of "practice" relates to specific, repeatable sequences of activities on which scientists rely in their daily work--things like the "plasmid prep" in molecular biology, discussed by Jordan and Lynch ( 1992). Unlike my generic sense of "practice," this one has a plural form; one can talk, for instance, of a distinct set of practices as characteristic of a given science or laboratory. And--this is the important point--practices, on my definition, fall into the sphere of culture, and the study of practices in their own right falls into the domain of what I just called cultural studies. In contrast, I am interested in practices not so much in themselves but inasmuch as they are among the resources for scientific practice and are transformed (or transformable) in practice, alongside all of the other components of scientific culture. I return to the practice/practices distinction in section 1.3, but for now the simple message is that thinking about science bifurcates in the


6. I cannot offer statistical evidence on this point, and it would not affect my arguments in the rest of the book if I were wrong. But still, my reading of the latest works of the science-studies authors I most admire tells me that they are largely engaged in projects different from my own. As an example of this, when I look back on the book I recently edited, Science as Practice and Culture (1992), I see now, as I did not then, that only Gooding's essay there shares my consistent interest in cultural transformation in time. The actor-network approach to science studies discussed in section 1.3 has been very important to my own thinking about the temporality of practice, but has largely been developed and appropriated by others in a cultural-studies mode, perhaps because many of its key terms--"centers of calculation," "obligatory passage points" as well as "network" itself--are terms of art in cultural mapping.


act of deciding whether "practice" has a plural or not. In this book, I take the latter fork.


Satisfied that the sequence of men led to nothing and that the sequence of their society could lead no further, while the mere sequence of tlme was artificial, and the sequence of thought was chaos, he turned at last to the sequence of force; and thus it happened that, after ten years' pursuit, he found himself Iying in the Gallery of Machines at the Great Exposition of 1900, his historical neck broken by the sudden irruption of forces totally new.

Henry Adams, The Education of Henry Adams

Before I get into the details of my understanding of scientific practice, I need to talk about the metaphysics that informs it. In particular, I want to contrast what I call the representational and performative idioms for thinking about science. The representational idiom casts science as, above all, an activity that seeks to represent nature, to produce knowledge that maps, mirrors, or corresponds to how the world really is. In so doing, it precipitates a characteristic set of fears about the adequacy of scientific representation that constitute the familiar philosophical problematics of realism and objectivity (which I discuss in chapter 6). Of course, within the traditionally restricted vision of science-asknowledge, the representational idiom is more or less obligatory--what else can one ask of knowledge other than whether it corresponds to its object?-but it has continued strongly to inflect our understandings of science even to the present. It has been taken to a new pitch of intensity, for example, in the reflexive approach to science studies developed in the works of Michael Mulkay (1985), Steve Woolgar (1988b, 1988c), and Malcolm Ashmore (1989). Reflexivity turns the usual fears concerning the adequacy of representation--the "methodological horrors," as Woolgar calls them--back upon science studies itself, but never doubts that the point of science and science studies is indeed representation.7

Within an expanded conception of scientific culture, however--one


7. Woolgar (1988a, 1992) makes clear the connection between the reflexive turn in science and that taking place more generally in the human sciences, especially in anthropology. The grip of the representational idiom is exemplified in Lynch's essay "Representation Is Overrated" (1993a). Lynch's conclusion is that we need to study representation differently, not that we should escape from the representational idiom.


that goes beyond science-as-knowledge, to include the material, social, and temporal dimensions of science--it becomes possible to imagine that science is not just about representation. And working through the studies discussed later in this book has convinced me that it is both possible and necessary to escape from the representational idiom if we are to get to grips with scientific practice. The point is this: Within the representational idiom, people and things tend to appear as shadows of themselves. Scientists figure as disembodied intellects making knowledge in a field of facts and observations (and language, as the reflexivists remind us). But there is quite another way of thinking about science. One can start from the idea that the world is filled not, in the first instance, with facts and observations, but with agency. The world, I want,to say, is continually doing things, things that bear upon us not as observation statements upon disembodied intellects but as forces upon material beings. Think of the weather. Winds, storms, droughts, floods, heat and cold--all of these engage with our bodies as well as our minds, often in life-threatening ways. The parts of the world that I know best are ones where one could not survive for any length of time without responding in a very direct way to such material agency--even in an English summer (never mind a midwestern winter) one would die quite quickly of exposure to the elements in the absence of clothing, buildings, heating, and whatever. Much of everyday life, I would say, has this character of coping with material agency, agency that comes at us from outside the human realm and that cannot be reduced to anything within that realm.S

My suggestion is that we should see science (and, of course, technology) as a continuation and extension of this business of coping with


8. I thank Malcolm Nicolson and Soraya de Chadarevian for emphasizing to me the relation between these remarks and the phenomenological tradition in Continental philosophy. I confess to little knowledge of the works of the masters, Husserl, Heidegger, and Merleau-Ponty, though Rouse (1985, 1986, 1987a) makes fascinating connections in seeking to inject these authors into contemporary Anglo- American philosophy of science. See also Rouse 1987b; Nicolson 1991; and note 27 below. I also thank Simon Schaffer and Steven Shapin for pressing me on my notion of material agency. They alerted me to the idea that concepts of agency and intentionality are bound together, and that therefore agency can properly be attributed only to human beings. I seek to clarify this issue below; in the meantime, I note that I am not alone in finding it useful and appropriate to speak of material agency. Thus, for example, Harre and Madden (1975) are happy enough to speak of agency in nature without ever dreaming of imputing intentionality to it, and Wise and Smith (1989-90, 419) quote William Whewell, writing in 1841 that "[i]n many cases the work to be done may be performed by various agencies; by men, by horses, by water, by wind, by steam."


material agency. And, further, we should see machines as central to how scientists do this. Scientists, as human agents, maneuver in a field of material agency, constructing machines that, as I shall say, variously capture, seduce, download, recruit, enroll, or materialize that agency, taming and domesticating it, putting it at our service, often in the accomplishment of tasks that are simply beyond the capacities of naked human minds and bodies, individually or collectively.9 A windmill grinds grain very much faster than a miller could do by hand; my television set shows me events distant m time and space that I could not otherwise hope to view; a machine tool cuts metal at a speed and with a precision that no one could otherwise hope to achieve.

These remarks, then, sketch out a basis for a performative image of science, in which science is regarded a field of powers, capacities, and performances, situated in machinic captures of material agency. And my aim in the rest of this book is to understand scientific practice within such a performative idiom. I can immediately add that thinking about material performativity does not imply that we have to forget about the representational aspects of science. Science is not just about making machines, and one cannot claim to have an analysis of science without offering an account of its conceptual and representational dimensions. Much of this book (chapters 3, 4, and 6) is therefore directly concerned with the production of scientific knowledge. The move to the performative idiom, however, does imply a certain strategy in thinking about scientific knowledge. It suggests that we should explore the many different and interesting ways in which knowledge is threaded through the machinic field of science. The performative idiom can thus include the concerns of the representational idiom; it is a rebalancing of our understanding of science away from a pure obsession with knowledge and toward a recognition of science's material powers.l° The machine, as I conceive it, is the balance point, liminal between the human and nonhuman worlds (and liminal, too, between the worlds of science, technology and society). I can also add that the move from a representational to a performative idiom in thinking about science has considerable historical plausibility. While the representational idiom might seem appropriate to


9. This formulation should make clear an alignment of my approach with that of the actor-network theorists, to whom I return below. The metaphor of domestication I take in particular from Callon (1986) and Latour (1987).

10. For an analysis that emphasizes the performative role in science of language itself see Rouse 1994; more generally, see also Smith 1993a.


classical naked-eye astronomy, most of the science of our day has evolved hand in hand with the machines that were at the heart of the industrial revolution of the eighteenth and nineteenth centuries (I return to this observation in chapter 7). Further, Martin Krieger's amazing philosophical anthropology of "doing physics" (1992) suggests that the performative idiom has a phenomenological warrant, too. Krieger's suggestion is that physicists take hold of the world as if it were a factory, a site of productive equipment that needs to be managed.

Outside the world of science studies, of course, the move from a representational to a performative idiom for thinking about science might seem obvious. As Krieger says, scientists themselves do, by and large, think about the world as a field of agency, and the machinic, material dimension of their dealings with it is hardly a secret to them. The scientific laity, too, I suspect, is less impressed by representations of nature than by the display of the machinic agency of science. The science pages of the popular press hardly distinguish between scientific theory and the latest high-tech gadgets, and devote more space to the latter. One might therefore have imagined that the representational idiom would be easily dislodged within science studies. Actually, though, there are many issues that need to be thought through in the move to a performative idiom, some of which we can now review.

First an issue that is not difficult. The weight of history is on the side of the representational idiom. Going back at least to the seventeenth century, philosophers have by and large sought to keep material agency out of their discussions of science.11 Even within philosophy of science, though, there is an alternative approach--exemplified in the work of Ian Hacking (1983), Nancy Cartwright (1983, 1989), Davis Baird and Alfred Nordmann (1994), and others--that points to the conclusion that a restriction to the representational idiom is by no means neces-


11. Ian Hacking (1983, 46) traces the philosophical aversion to speech about agency and causes in nature back to Newton's work on gravitation: Newton provided an explanation for the motion of the planets that did not, in the terms of his age, include an account of the detailed causes of that motion. Both Bloor (1983, 152-55) and Shapin (1982, 181-84) review the literature on arguments for the passivity of matter put forward in seventeenth-century England, relating them to their social, moral, religious, and political context (without remarking that SSK itself reproduces the same image of matter; see below). Philosophical arguments for the performative rather than the representational idiom often identify David Hume as a key figure in the effacement of material agency from philosophical discourse: see Cartwright 1989, 2, for example (and, from a different angle, Husserl 1970, 84-90).


sary.l2 As just noted, the world of science happens to be quite evidently and amply stocked with material agents. Hacking's well-known slogan (1983, 23)--"if you can spray them then they are real"--refers, for example, to the use of radioactive sources to change the electric charges of samples in quark-search experiments.l3 I am not persuaded of Hacking's realism, but such radioactive sources are certainly instances of what I have in mind as material agency--they are objects that do things in the world. So the sheer weight of history and custom behind the representational idiom in philosophy need not deter us from moving toward a performative one. But there are more difficult problems that should give us pause, problems that are better approached from a sociological rather than a philosophical perspective.14


The great achievement of SSK was to bring the human and social dimensions of science to the fore. SSK, one can say, thematized the role of human agency in science. It thus partially displaced the representational idiom by seeing the production, evaluation, and use of scientific knowledge as structured by the interests of and constraints upon real human agents. Scientific beliefs, according to SSK, are to be sociologically accounted for in just those terms. Thus, in his classic study of the phrenology debates in early-nineteenth-century Edinburgh, Steven Shapin (1979) argues that the differing accounts of the brain produced by the competing parties have to be understood not in terms of their correspondence, or lack of it, to how the brain really is, but in relation to the divergent social interests of the phrenologists and their establishment critics: the phrenologists wanted to reform society in ways that the establishment resisted. On this view, then, scientific representations of na-


12. See also Bhaskar 1975; Chalmers 1992, Harre and Madden 1975, and Mellor 1974. I thank Peter Lipton for introducing me to this line of philosophical thought.

13. Hacking formulates this slogan while discussing the work of William Fairbank that figures in section 6.6 below.

14. In philosophy, the best-known general critique of representationalism is Rorty 1979. Hacking 1983 is the opening salvo in philosophy of science, but the most sustained encounter with, and critique of, representationalism is Rouse forthcoming. Rorty's favored alternative is the move to a conversational idiom. The shift I recommend is more like Hacking's move from representing to intervening. Rouse recommends a cultural-studies approach to science.


ture have to be understood in terms of particular configurations of human agency.

Now, there is no denying the fruitfulness of the eruption of SSK in science studies. It has irreversibly transformed the field for the better. But there is also no denying that it stands in the way of developing a fully performative understanding of science. The problem is that SSK makes it impossible to take material agency seriously. The other side of SSK's focus on human agency is precisely the invisibility of material agency. SSK, that is, is happy to talk of accounts of material agency as components of scientific knowledge, but it insists that such accounts be analyzed, like any other component, by referring them back to some field of human agency. The question for SSK is always, why does some community of scientists have this account of material agency? and any appeal to material agency and performativity as part of the answer is ruled out from the start.l5 SSK's account of science is, then, at most semiperformative.

Now, why should SSK adopt this asymmetric stance concerning material and human agency? Part of the answer, I think, is that it has unreflectively taken over the representational idiom from traditional philosophy of science.l6 But part of the answer is more interesting. Within SSK, one can generate arguments that point up real difficulties in moving to a fully performative idiom that acknowledges both material and human agency. And I want now to review a debate that crystallizes those difficulties, the so-called chicken debate, launched by Harry Collins and Steven Yearley's essay (1992a), "Epistemological Chicken." Collins and Yearley's target is the actor-network approach developed in the works


15. I confess that I argued along these lines in Pickering 1984c, chap. 1; 1984a. For an extension of the standard SSK perspective into technology studies, see Pinch and Bijker 1984.

16. Thus, perhaps, Bloor (1991, 158) regrets the absence of material agency in his canonical articulation of the SSK position: "The shortcomings of the views developed here are, no doubt, legion. The one I feel most keenly is that, whilst I have stressed the materialist character of the sociological approach, still the materialism tends to be passive rather than active. It cannot, I hope, be said to be totally undialectical, but without doubt it represents knowledge as theory rather than practice." Interestingly, Bloor goes on to compare his own position with the philosophy of Karl Popper, who "makes science a matter of pure theory rather than reliable technique." (I thank David Bloor for drawing my attention to these quotations.) In contrast, Barnes (1991, 331) is forthright in his rejection of a role for material agency in sociological accounting: "Reality will tolerate alternative descriptions without protest. We may say what we will of it, and it will not disagree. Sociologists of knowledge rightly reject realist epistemologies that empower reality."


of Michel Callon, Bruno Latour, end John Law.l7 Since the early 1980s, the actor-network has, I believe, shown the way toward a fully performative understanding of science, and thus a discussion of Callon and Latour's response (1992) to Collins and Yearley can serve two purposes here. On the one hand, it offers me the occasion to emphasize that I have learned an enormous amount from the actor- network. Its traces will be evident throughout this book, in the text and footnotes, and I would happy if the book were read as an attempt at a constructive dialogue with the actor-network. On the other hand, a discussion of the points at which my analysis diverges from that of the actor-network will enable me to set out my own thoughts on time, agency, and practice.18

The basic metaphysics of the actor-network is that we should think of science (and technology and society) as a field of human and nonhuman (material) agency. Human and nonhuman agents are associated with one another in networks, and evolve together within those networks. The actor- network picture is thus symmetrical with respect to human and nonhuman agency. Neither is reduced to the other; each is constitutive of science; therefore we need to think about both at once. This "extended symmetry" of humans and nonhumans is the target of Collins


17. For recent important presentations and developments of actor-network theory, see Callon 1994; Latour 1987,1988a; and Law 1993. Rouse 1987b is an excellent philosophical discussion and extension of the actor-network approach. I should note that Collins and Yearley's essay actually has two targets: the second is Steve Woolgar's reflexivity program. As indicated earlier, I understand reflexivity as an intensification of the representational idiom in science studies, and since I want to escape from that idiom, I will not discuss Collins and Yearley's critique of it here. One point is, however, worth clarifying. Reflexivity is, like the actor-network, symmetric about human and material agency, but in a negative rather than a positive way. Reflexivity adds to SSK's deconstruction of scientists' accounts of material agency a deconstruction of SSK's accounts of the human agency of scientists. I also note below that, when the actor-network makes its semiotic move, it appears to circle back toward reflexivity. Woolgar 1992 is his response to Collins and Yearley's critique; Collins and Yearley 1992b is their response to Callon and Latour's response (1992) to Collins and Yearley 1992a.

18. One shortcoming of this expository strategy is that it ties my analysis of practice more closely to material agency than is necessary; as explained in section 1.5, my overall analysis of practice is applicable even to situations where material agency is absent. A more specific point of divergence from the actor-network than those that follow can also be noted here, concerning the "network" metaphor itself. While it performs useful work in the classic actor-network studies, I have not found it illuminating in connection with my own studies, and it plays no role in the following chapters. Mol and Law (1994) argue for a more complex "social topology" of "regions," "networks," and "fluids," without suggesting that this is an exhaustive list. I deploy some further topological concepts in section 7.4.


and Yearley's critique, which hinges upon the following dilemma. As analysts, Collins and Yearley suggest, we have just two alternatives. We can see scientists as producing accounts of material agency, in which case these accounts fall into the domain of scientific knowledge and should be analyzed sociologically as the products of human agents. This is the standard SSK position that Collins and Yearley want to defend.l9 Or we can try to take material agency seriously, on its own terms-- but then we yield up our analytic authority to the scientists themselves. Scientists, not sociologists, have the instruments and conceptual apparatus required to tell us what material agency really is. The upshot of this dilemma therefore seems to be that any sociologist with a shred of self-respect had better stick to an SSK-style analysis of scientific accounting for material agency, and had better not incorporate material agency per se into her interpretive schemes. The alternative seems to be a simple rehearsal of scientific/engineering accounts of how the world is, from which the human and the social simply disappear. One can have human or nonhuman agency, but not both. This is the basic problem in moving to a fully performative idiom, as seen from the perspective of SSK.

On behalf of actor-network theory, Callon and Latour (1992) reply to Collins and Yearley that they do not see things quite this way, and neither do I, though here our positions diverge as discussed below. Callon and Latour reject the prongs of Collins and Yearley's dilemma. They insist, rightly I think, that there are not just two alternatives in the treatment of nonhuman agency. And their position, if I have understood it correctly, is this. We should not see nonhuman agency in the terms offered us by scientists or by SSK. Instead we should think semiotically. Semiotics, the science of signs, teaches us how to think symmetrically about human and nonhuman agents. In texts, agents (actors, actants) are continually coming into being, fading away, moving around, changing places with one another, and so on. Importantly, their status can easily make the transit between being real entities and social constructs, and back again. Semiotics thus offers us a way of avoiding the horns of


19. A certain incongruity will become apparent as this chapter and the book goes on. I find much of Collins's writing, both before and after the chicken debate, and even the later passages of Collins and Yearley's second contribution to it (1992b), to be more consistent with the actor- network approach and the analysis of practice that I want to develop than with the pure-line SSK of Collins and Yearley 1992a. I have no compelling analysis of this inconsistency; perhaps Collins's humanism (see below) is in tension with his technical analyses. To put the point more positively, I have learned very much from Collins in developing my own account of practice.


Collins and Yearley's dilemma: the agencies we speak about are semiotic ones, not confined to the rigid categories that SSK and engineering impose.

This is a clever and ingenious response, but it brings with it two different kinds of problem. One concerns human agency, and I return to it below. The other concerns material agency. From my point of view, the most attractive feature of the actor-network approach is precisely that its acknowledgment of material agency can help us to escape from the spell of representation. It points a way to a thoroughgoing shift into the performative idiom. And from this perspective, the appeal to semiotics in the face of Collins and Yearley's dilemma looks like a kind of retreat, a return to the world of texts and representations that one does not wish to make.20 Fortunately, as I shall now explain briefly in anticipation of the empirical examples to follow, there is another way of steering our way around the dilemma.


20. Semiotics cannot be the whole story about the actor-network understanding of nonhuman agency, but it has been a central theme since Latour and Woolgar's (1986) emphasis on scientific instruments as "inscription devices," and seems invariably to be invoked under pressure. Thus, in response to Collins and Yearley's argument, Callon and Latour draw a diagram (1992, 349, fig. 12.3) in which Collins and Yearley's dilemma is represented as a horizontal nature/society axis, to which Callon and Latour add a vertical axis on which they comment (350): "The vertical axis, however, is centered on the very activity of shifting out agencies--which is, by the way, the semiotic definition of an actant devoid of its logo- and anthropocentric connotations." Likewise, in response to Schaffer's argument (1991) that Latour (1988b) depends on an illegitimate hylozoism--that is, an unjustified imputation of agency to the nonhuman realm--Latour (1992a) offered a reading of a single memoir by Pasteur. It seems clear that Latour's route to nonhuman agency in this instance is via texts in the most literal sense. (I have only an incomplete draft of Latour 1992a, but I attended his seminar presentation of the paper--Cambridge, October 1992--and subsequent conversations with him confirmed this judgment. The substance of Latour's talk has since been published (1992b), not as a response to Schaffer but as an exemplification of the semiotic method.) In a similar vein, Callon writes: "Sociology is simply an extension of the science of inscriptions. Now it should broaden its scope to include not only actors but the intermediaries through which they speak . . . the social can be read in the inscriptions that mark the intermediaries" (1991, 140). The position that I take below is closer to that developed earlier by John Law in, for example, Law 1987, where, without any detours through semiotics, he invokes natural forces as part of an actor-network account of the Portuguese maritime expansion. One can also note that even when Latour claims to be doing semiotics he often seems to forget, and to speak directly about laboratory practices (rather than textual accounts thereof): see, for example, key passages in Latour 1992b. More formally, Latour's recent writings introduce a peculiar quasi-semiotic operator called "shifting down," which moves between texts and the material world--between, to mention his own example, textual representations of air travel and the real thing (1993a, 16). I think that the semiotic exposition of the actor-network starts to unravel at this point.


My alternative to the invocation of an atemporal science of signs is to think carefully about time. We can take material agency in science just as seriously as SSK takes human agency, and still avoid Collins and Yearley's dilemma, if we note that the former is temporally emergent in practice. The contours of material agency are never decisively known in advance, scientists continually have to explore them in their work, problems always arise and have to be solved in the development of, say, new machines.21 And such solutions--if they are found at all--take the form, at minimum, of a kind of delicate material positioning or tuning, where I use "tuning" in the sense of tuning a radio set or car engine, with the caveat that the character of the "signal" is not known in advance in scientific research.22 Thus, if we agree that, as already stipulated, we are interested in achieving a real-time understanding of scientific practice, then it is clear that the scientist is in no better a position than the sociologist when it comes to material agency. No one knows in advance the shape of future machines and what they will do, but we can track the


21. One can make a connection to early pragmatist philosophy here. Discussing the work of Charles Sanders Peirce, Cohen (1923, xix n. 11) explains that "Peirce's tychism is indebted to [Chauncey] Wright's doctrine of accidents end 'cosmic weather,' a doctrine which maintained against LaPlace that a mind knowing nature from moment to moment is bound to encounter genuine novelty in phenomena, which no amount of knowledge would enable us to foresee." The same doctrine is expressed in William James's well-known sentiment that "[e]xperience, as we know, has ways of boiling over, and making us correct our present formulas" ([1907, 1909] 1978, 106).

22. The notion of tuning is close to Knorr-Cetina's notion (1981) of tinkering, except that the former immediately invokes the otherness of material (and other kinds of) agency. As noted in chapter 3, it was in thinking about the tuning of material apparatus that I became convinced of the need to include material agency in my analysis of scientific practice. Fleck ([1935] 1979) discusses the tuning (in my terms) of the Wassermann reaction as a test for syphilis. He notes that "[d]uring the initial experiments it produced barely 15-20 percent positive results in cases of confirmed syphilis" (72), but that after a period of collective development of the detailed performance of the reaction the success rate rose to 70-90 percent. Collins's account (1992, chap. 3) of the work of building a TEA laser can likewise be read as an ethnographic exemplification of the tuning of a material instrument. For more on the tuning of experimental devices in science, see Pickering (1984b, 14 and passim) and the works cited there. Note that it is crucial that we are considering scientific practice as the work of extending, rather than reproducing; scientific culture-- in the sense of building new machines and so on. I argue that material agency is temporally emergent in relation to practice so conceived. Whether material agency per se is temporally emergent is another matter. The relative reliability of certain machines--the fact that some magnets, cars, and TVs perform the same functions day after day, for example--indicates that some aspects of material agency evolve at most slowly on the time scale of human affairs. I return to this topic in section 7.6; and I thank Adrian Wilson for prompting me to think about it in the first place.


process of establishing that shape without returning to the SSK position that only human agency is involved in it. Of course, after the fact, scientists often offer persuasive technical accounts of why the machinic field of science has developed in specific ways. But, as I said earlier, for the purposes of real-time accounting, the substance of such retrospective accounts is one aspect of what needs to be analyzed; it would make no sense to bow to the scientists and incorporate their retrospection as part of our explanation. This is my basic thought on how to think about the role of nonhuman agency in scientific practice.

Now for the second difficulty with the actor-network's semiotic move, the one that centers on human agency. Semiotics imposes an exact symmetry amounting to an equivalence and interchangeability between the human and material realms. Semiotically, as the actor-network insists, there is no difference between human and nonhuman agents. Semiotically, human and nonhuman agency can be continuously transformed into one another and substituted for one another. I am not alone in thinking that there are serious problems with these ideas when it comes to the analysis of science. As agents, we humans seem to be importantly different from nonhuman agents like the weather, television sets, or particle accelerators. Certainly, for example, even the proponents of the actor-network seem to have little difficulty separating people and things when they write about science, as Yves Gingras (1994) has forcefully pointed out. More precisely, the idea that, say, human beings can be substituted for machines (and vice versa) seems to me a mistake. I find it hard to imagine any combination of naked human minds and bodies that could substitute for a telescope, never mind an electron microscope, or for a machine tool, or for an atom bomb (or for penicillin, heroin, . . .). Semiotically, these things can be made equivalent; in practice they are not.23 But still, I believe that the actor-network is onto something in its extended symmetry, actually two things. One is that there exist important parallels between human and material agency, concerning both their repetitive quality and their temporal emergence; the other is that a constitutive intertwining exists between material and human agency. Or so my studies of practice convince me. I now want to explain these remarks, in a performative rather than a semiotic idiom.


23. The actor-network speaks of "delegating" human performances to machines; my suggestion is that the putative symmetry of this operation often breaks down when one tries to imagine delegating machinic functions back to humans.


When I have finished, most of the important elements of my understanding of practice will be in place, and I can pull them together in the next section.

We can start with scientific culture, as I defined it earlier, rather than with practice. Think of the field of machines that constitute the established material performativity of science at any given time. This machinic field does not exist in a human vacuum. Though the machines and instruments of science often display superhuman capacities, their performativity is nevertheless enveloped by the human realm. It is enveloped by human practices (recall the discussion of section 1.1)--by the gestures, skills, and whatever required to set machines in motion and to channel and exploit their power. Though it is not strictly about machines, one can think, for example, of Fleck's classic discussion of the history of the Wassermann reaction as a test for syphilis ([1935] 1979), in which he emphasizes the importance of the establishment of a community of practitioners competent to carry through the test. In the absence of such a community and its specific human performances, the test would not exist. Further, just as the performativity of machines is repetitive--my computer displays the same powers day after day--so is the human performativity that envelops them. The field of practices is routinized and disciplined, machinelike, as Collins puts it.24 The proper performance of the Wassermann reaction requires adherence to a standardized sequence of gestures and manipulations; around machines, we act like machines. In these simple senses, then, there is, if not a perfect interchangeability, a very important degree of symmetry and interconnection between human agency and material agency: as respectively disciplined in practices and as captured in machines, they are both repetitive and machinelike and they collaborate in performances.

Now we can turn to practice, to the extension of scientific culture, and especially to the extension of the machinic field of science. I have already described this as involving a process of tuning, and the key point about tuning in the present context is that it works both ways, on human as well as nonhuman agency. Just as the material contours and performativity of new machines have to be found out in the real time of practice, so too do the human skills, gestures, and practices that will envelop them. It would make no sense, for example, to try to imagine


24. See Collins 1990, 1994; Collins and Kusch forthcoming a, forthcoming b. Collins speaks interchangeably of "machine-like" and "behaviour-specific" action; I prefer the former locution, for obvious reasons.


some community developing the skills implicit in the execution of the Wassermann reaction in the absence of the material technologies of the procedure. Gestures, skills, and so on--all of these aspects of disciplined human agency come together with the machines that they set in motion and exploit.25 Here, then, we can note further degrees of symmetry between human and material agency. Just as material agency is temporally emergent in practice, so, necessarily, is disciplined human agency. And, furthermore, while the two are not continuously deformable into one another--one cannot substitute human beings for the reagents of the Wassermann reaction--they are intimately connected with one another, reciprocally and emergently defining and sustaining each other. Disciplined human agency and captured material agency are, as I say, constitutively intertwined; they are interactively stabilized (Pickering 1989c).26

There are other aspects of human agency besides machinelike practices that we will need to discuss as we go along; I will later talk about the scale and social relations of human actors and argue that these too are tuned and interactively stabilized in practice. For the moment, though, I want to discuss an aspect in which the symmetry between human and material agency appears to break down. I want to talk about intentionality--a term I use in an everyday sense to point to the fact that scientific practice is typically organized around specific plans and goals.27 I find that I cannot make sense of the studies that follow without reference to the intentions of scientists, to their goals and plans, though I do not find it necessary to have insight into the intentions of things. One has to recognize that scientists usually work with some future destination in view, whereas it does not help at all to think about machines in the same way. Human intentionality, then, appears to have no coun-


25. This is Knorr-Cetina's point (1992, 1994) when she writes about the laboratory as the home and site of mutual reconfiguration of "enhanced nature" and "enhanced agents"--though the implicit contrast between "nature" and "agents" seems misplaced to me. Polanyi 1958 is an early classic on skill and "tacit knowledge" in science.

26. Or, as Callon and Latour (1992) put it (following Engels), coproduced, in what John Law (1987) calls a process of heterogeneous engineering.

27. Piet Hut has encouraged me to note that my usage of "intentionality" is not the technical usage found in phenomenological philosophy, where it evokes the idea that consciousness is always consciousness of something. On the other hand, there seem to be many points of contact between my overall analysis and Husserl 1970. The discussion that follows here, for example, could be taken as an elaboration of Husserl's remark that ''[w]orld is the universal field into which all our acts, whether of experiencing, of knowing, or of outward action, are directed. From this field, or from objects in each case already given, come all affections, transforming themselves in each case into actions" (1970, 144).


terpart in the material realm. But still, I want to stress here the significant parallels and intertwinings between the intentional structure of human action and material agency. Especially I want to stress the temporal emergence of plans and goals and their transformability in encounters with material agency.25

We could start from the idea that perhaps a perfect symmetry does obtain between human and nonhuman agency, even if human intentionality is included in the picture. At some deep level, perhaps the vectors of human practice are just as temporally emergent from moment to moment and situation to situation as is material agency. Just as we do not know what this new machine will do, we do not know what other people, or even we ourselves, will do next. Think of the characters in J. G. Ballard's stories--always acting, but never with any vision of where their actions will take them. From this perspective, there would be an exact parallel between the intentional structure of human agency and material agency prior to any machinic capture: both are wild and undomesticated. In the studies that follow, however, human goals and purposes do not seem to be as wild as all that. They appear as already partially tamed, already on the way to being brought to heel by the cultures in which they are situated, as I can now explain.

In trying to understand the intentionality of scientific practice, it is important to continue to pay attention to time. Sometimes, at least, we humans live in time in a particular way. Unlike DNA double helixes or stereo systems, we construct goals that refer to presently nonexistent future states and then seek to bring them about.29 We aim to build a new kind of machine, say, that we hope will display certain powers. This


28. Bruno Latour and Barbara Herrnstein Smith have criticized my views on intentionality, though I cannot discern any inconsistency between the position developed here and theirs. On Latour, see note 29 below; on Smith, see her words at the beginning of section 2. 3. Two further points are worth noting. First, the question of intentionality, as I have defined it, simply does not arise in synchronous studies of culture (including studies of practices). Second, the following argument states, and the studies discussed in part 1 exemplify, my conviction that the plans and goals of practice are not derivable from existing culture. No static account--of scientific reason as "problem-solving," or of "constraints" upon practice, or whatever--can explain why particular actors set off along particular vectors of cultural extension. Something more is needed, which my account seeks to supply.

29. Ashmore 1993 is a brave and amusing, but not very persuasive, attempt to attribute intentionality to a material agent, namely, a "catflap." The other way to preserve a symmetry between nonhuman and human agency is, of course, to deny intentionality to humans. This, I think, is John Law's strategy (1993) in describing human agents as "network effects," "performed" by organizational narratives or "myths" (note the return from


extended temporal sweep of human agency is, for me, a respect in which the symmetry between human and material agency breaks down, and without marking this difference I cannot make sense of the case studies that follow. But having marked it, more needs to be said. If one defines intentionality in terms of human plans and goals, the questions that arise concern the origin and substance of such goals, and here traditional studies of science already offer an answer. The goals of scientific practice are imaginatively transformed versions of its present. The future states of scientific culture at which practice aims are constructed from existing culture in a process of modelling (metaphor, analogy).30 This, stated in a few words, is my basic idea of how existing culture predisciplines the extended temporality of human intentionality. I will exemplify and elaborate this idea later, but here I want just to make three comments that will connect it to my present theme of the parallels between and intertwining of human and material agency.

First, the predisciplining of intent by existing culture is only partial. Modelling is an open-ended process with no determinate destination. From a given model--say a particular functioning machine--an indefinite number of future variants can be constructed. Nothing about the model itself fixes which of them will figure as the goal for a particular passage of practice (Barnes 1982 expresses this point very well).31 There is no algorithm that determines the vectors of cultural extension, which is as much as to say that the goals of scientific practice emerge in the


agency to textuality, paralleling the semiotic turn we have been discussing). It is worth noting that Latour often seems to have a pretty distinct notion of human intentionality. While the early image in Latour and Woolgar 1986 of an "agonistic" war of all against all--a general intention to dominate in battle--can be carried over symmetrically to nonhuman agents or to networks as wholes (Callon and Latour 1981), I do not think that the same can be said of Latour's later discussion of "translating interests" (1987, 108-21). This latter seems to me to be applicable only to intentional human agents acting on other intentional human agents. For a thoughtful review of sociological understandings of intentionality, see Lynch 1992c; for a survey of recent thinking in the social sciences on human and nonhuman agency, see the contributions to Ashmore, Wooffitt, and Harding 1994.

30. I take the idea of modelling from discussions in history and philosophy of science of the role of metaphor and analogy in theory development. I prefer to speak of modelling since I want to apply the idea to the material and social, as well as the conceptual, aspects of scientific culture, while metaphor and analogy are usually taken as having textual referents. Modelling, in Kuhnian terms, is developing an exemplar. For access to the relevant literature, see, for example, Barnes 1982; Bloor 1991; Gooding 1990; Hesse 1966; Knorr-Cetina 1981; Kuhn 1970; and Pickering 1981c, 1984b.

31. Such open-endedness in human agency seems to me a necessary counterpart to the emergent quality of material agency; it is what makes it possible to bring the two into relation with one another.


real time of practice. Existing culture, to appropriate Michel Foucault's phrase (1972), is literally the surface of emergence for the intentional structure of human agency. Even in the domain of intentionality, then, this temporally emergent quality constitutes an important parallel between human and material agency. Second, if the field of existing machines serves as a surface of emergence for the goals of scientific practice, then human intentions are bound up and intertwined (in many ways) with prior captures of material agency in the reciprocal tuning of machines and disciplined human performances. The world of intentionality is, then, constitutively engaged with the world of material agency, even if the one cannot be substituted for the other. And third, I can note that, as I conceive it, tuning can also transform the goals of scientific practice. Scientists do not simply fix their goals once and for all and stick to them, come what may.32 In the struggles with material agency that I call tuning, plans and goals too are at stake and liable to revision. And thus the intentional character of human agency has a further aspect of temporal emergence, being reconfigured itself in the real time of practice, as well as a further aspect of intertwining with material agency, being reciprocally redefined with the contours of material agency in tuning.33

Thus my preliminary analysis of the intentional structure of human agency. I find myself forced to regard the latter as differing from nonhuman agency in its temporal structure, through its orientation to goals located in the future. But still I need to emphasize that such goals should be seen as in the plane of practice (to paraphrase my opening quotation from Deleuze and Parnet) rather than as controlling practice from without. Goals are temporally emergent from culture (including machines and their material performativities) and can themselves be transformed in, and as an integral part of, real-time practice, which includes sensitive encounters with material agency.34


32. Fuller (1992) thinks that scientists should be taken to task for this; to the contrary, I take it to be central to scientific creativity.

33. Collins seems to me to be the only writer within the SSK canon who is directly concerned with the temporal emergence of human agency. See Collins 1992 and my essay review, 1987. More generally, recognition of emergence in human agency aligns my position with symbolic interactionist, ethnomethodological, and pragmatist sociologies: Denzin 1992 surveys the history of symbolic interactionism up to the present, stressing its links with pragmatism and ethnomethodology; Lynch 1992a and 1993b are good entry points for ethnomethodological studies of science; for access to pragmatist studies of science and technology, see Star 1991b, 1992; and Fujimura 1992. Laurel Graham and John Law have long encouraged me to think about the relation between my studies of science and symbolic interactionism; I regret that I did not follow their suggestions earlier.



"Do not worry if you think it is dark," he said to me, "because I am going to light the light and then mangle it for diversion and also for scientific truth."

Flann O'Brien, The Third Policeman

As promised, most of the elements of my understanding of scientific practice have been laid out in picking my way through the chicken debate. Now I can sum them up and begin to elaborate upon them. My basic image of science is a performative one, in which the performances--the doings--of human and material agency come to the fore. Scientists are human agents in a field of material agency which they struggle to capture in machines. Further, human and material agency are reciprocally and emergently intertwined in this struggle. Their contours emerge in the temporality of practice and are definitional of and sustain one another. Existing culture constitutes the surface of emergence for the intentional structure of scientific practice, and such practice consists in the reciprocal tuning of human and material agency, tuning that can itself reconfigure human intentions. The upshot of this process is, on occasion, the reconfiguration and extension of scientific culture--the construction and interactive stabilization of new machines and the disciplined human performances and relations that accompany them.

This is the skeleton of my overall scheme. Now I want to expand it a little by returning once more to the idea of tuning. This will bring us to the mangle. I find tuning a perceptive metaphor, but in the detailed analysis of practice it helps to decompose it along the following lines. Tuning in goal- oriented practice takes the form, I think, of a dance of agency. As active, intentional beings, scientists tentatively construct some new machine. They then adopt a passive role, monitoring the performance of the machine to see whatever capture of material agency it might effect. Symmetrically, this period of human passivity is the period in which


34. It might help to clarify my position if I were to endorse some sentiments expressed by Lucy Suchman against an alternative view of human intentionality. She writes: "[O]ur actions, while systematic, are never planned in the strong sense that cognitive science would have it. Plans are best viewed as a weak resource for what is primarily ad hoc activity. . . Stated in advance, plans are necessarily vague, insofar as they must accommodate the unforeseeable contingencies of particular situations" (1987, ix). And, "rather than subsume the details of action under the study of plans, [we should think about how] plans are subsumed by the larger problem of situated action" (1987, 50). I agree.


material agency actively manifests itself. Does the machine perform as intended? Has an intended capture of agency been effected? Typically the answer is no, in which case the response is another reversal of roles: human agency is once more active in a revision of modelling vectors, followed by another bout of human passivity and material performance, and so on.35 The dance of agency, seen asymmetrically from the human end, thus takes the form of a dialectic of resistance and accommodation, where resistance denotes the failure to achieve an intended capture of agency in practice, and accommodation an active human strategy of response to resistance, which can include revisions to goals and intentions as well as to the material form of the machine in question and to the human frame of gestures and social relations that surround it.36

The practical, goal-oriented and goal-revising dialectic of resistance


35. Fleck ([1935] 1979) is the inspiration for the active/passive distinction made here. Fleck also uses the metaphor of "tuning," giving it a specifically musical sense: "Wassermann heard the tune that hummed in his mind but was not audible to those not involved. He and his co-workers listened and 'tuned' their 'sets' until these became selective. The melody could then be heard even by unbiased persons who were not involved" (86). Later, Fleck socializes the musical metaphor: "The matching of all the five required reagents, so as to maximize the effect of the reactions and ensure that the results a re as clear as possible, requires experience. Even quasi-orchestral practice is needed if, as is usual, the test is performed by a team" (97). Likewise, Lynch argues that to try to make sense of activities in a student laboratory without reference to the events that the students are witnessing through microscopes "would be analogous to analyzing an audiovisual recording of a symphony with the sound turned off" (1991b, 52). These musical figures can be understood as part of an attempt to escape from traditional visual/representational understandings of science, but they remain, in the end, asymmetric: the metaphor should be dancing (with partners), not listening or even performing. As usual, Latour is more surefooted: "An experiment shifts out action from one frame of reference to another. Who is acting in this experiment? Pasteur and his yeast. More exactly, Pasteur acts so that the yeast acts alone. . . he creates a scene in which he does not have to create anything. He develops gestures, glassware, protocols, so that the entity, once shifted out, becomes automatic and autonomous. . . Who is doing the acting in the new medium of culture? Pasteur, since he sprinkles, and boils, and filters, and sees. The lactic acid yeast, since it grows fast, uses up its food, gains in power . . ., and enters into competition with other similar beings growing like plants in the same soil. If I ignore Pasteur's work, I fall into the pitfalls of realism . . . if we ignore the lactic acid . . . [w]e fall into the other pit . . . of social constructivism, forced to ignore the role of nonhumans . . . We do not have to choose between two accounts of scientific work, since this very scientific work aims at building a scene in which scientists do not do any work" (1992b, 141-44). (These sentiments do not seem very semiotic to me.)

36. I thank Yves Gingras for pointing out the close relation between what I call the dialectic of resistance and accommodation and Jean Piaget's account ( 1985) of intellectual development. "Accommodation" is one of Piaget's key terms, often understood in respect of "resistances" and "obstacles," and his notion of "equilibration" is close to my idea of interactive stabilization. Piaget's empirical interests are different from mine, and his ideas are elaborated more in a representational than in a performative idiom, but it would be


and accommodation is, as far as I can make out, a general feature of scientific practice. And it is, in the first instance, what I call the mangle of practice, or just the mangle. I find "mangle" a convenient and suggestive shorthand for the dialectic because, for me, it conjures up the image of the unpredictable transformations worked upon whatever gets fed into the old-fashioned device of the same name used to squeeze the water out of the washing. It draws attention to the emergently intertwined delineation and reconfiguration of machinic captures and human intentions, practices, and so on. The word "mangle" can also be used appropriately in other ways, for instance as a verb. Thus I say that the contours of material and social agency are mangled in practice, meaning emergently transformed and delineated in the dialectic of resistance and accommodation.37 In a broader sense, though, throughout the book I will take the mangle to refer not just to this dialectic and the transformations it effects, but to the overall scheme, as I just called it, to the overarching image of practice that encompasses the dialectic--to the worldview or metaphysics, if you like, which sees science as an evolving field of human and material agencies reciprocally engaged in a play of resistance and accommodation in which the former seeks to capture the latter.

Thus the mangle, which in its several senses is the subject of all that follows. To get much further, we really need to turn to some examples, but before we do, one last remark might help. I do not think that the workings of the mangle are hard to grasp in any particular instance, but two aspects of the overall analysis are. One is the concept of temporal


interesting to work through the extension of his scheme to scientific practice. In Piaget's terms (1985, 17), my concern is with "homeorhesis" rather than with "homeostasis."

37. If pressed too hard, the mangle metaphor quickly breaks down. A real mangle leaves the list of clothing unchanged--"shirts in, shirts out"--which is too conservative an image for the constructive aspect of scientific practice. "Mangling" also carries connotations of mutilation and dismemberment--"my teddy bear was terminally mangled in a traffic accident"--which carry one directly away from this constructive aspect. There is little to be done about this; I can think of no more appropriate word, one has simply to try to take the metaphor seriously enough but not too seriously. (Alternatively, one could think of the productive and transformative magic mangle in Flann O'Brien's The Third Policeman.) I thank, among others, Mike Lynch, Ted O'Leary, and Allan Megill for warning me of potential difficulties with mangle talk and encouraging me to explain it more fully. (It turns out that there are even disputes about what a mangle is as domestic technology. Natalie Davis and Frederick Suppe tell me that their mothers' mangles were devices to speed up the ironing. This might suggest that the meaning of mangle differs between England, where I learned the term, and the US, but Richard Burian offered to take me into the depths of West Virginia and show me mangles, in my sense, still in use.)


emergence. What emergence entails is exemplified in the earlier brief discussion of material agency. In advance, we have no idea what precise collection of parts will constitute a working machine, nor do we have any idea of what its precise powers will be. There is no thread in the present that we can hang onto which determines the outcome of cultural extension. We just have to find out, in practice, by passing through the mangle, how the next capture of material agency is to be made and what it will look like. Captures and their properties in this sense just happen. This is my basic sense of emergence, a sense of brute chance, happening in time--and it is offensive to some deeply ingrained patterns of thought. The latter look for explanations--and the closer to the causal, mechanical explanations of classical physics the better--while it seems to me that in the analysis of real-time practice, in certain respects at least, none can be given. I can do nothing about this, but it is best to be clear on this point from the start. The world of the mangle lacks the comforting causality of traditional physics or engineering, or of sociology for that matter, with its traditional repertoire of enduring causes (interests) and constraints. I must add, though, that in my analysis brute contingency is constitutively interwoven into a pattern that we can grasp and understand, and which, as far as I am concerned, does explain what is going on. That explanation is what my analysis of goal formation as modelling, the dance of agency, and the dialectic of resistance and accommodation is intended to accomplish. The pattern repeats itself endlessly, but the substance of resistance and accommodation continually emerges unpredictably within it.38


38. At stake here is the question of what explanation and understanding consist in, a topic of interesting debate in contemporary philosophy. Mangle-ish explanation of scientific practice has much the same explanatory character as that ascribed to chaos theory in the natural sciences by Stephen Kellert: in contrast to traditional scientific explanations, "chaos theory does not provide predictions of quantitative detail but of qualitative features; it does not reveal hidden causal processes but displays geometric mechanisms; and it does not yield lawlike necessity but reveals patterns. . . Chaos theory takes up [an] emphasis on finding patterns and connections, while jettisoning the requirement that the patterns must yield necessity in a detailed and deterministic sense" (1993, 105, 112). One might argue that this just shows that chaos theory (and the mangle) explains nothing, but Kellert notes on behalf of chaos theory--as I do later for the mangle--that there are many natural phenomena on which traditional explanatory schemata find no purchase: "an account of scientific understanding as the disclosure of hidden causal processes is not only inadequate for the biological and social sciences . . . but it is inadequate for the physics of nonlinear systems as well" (105). Kellert also (113) draws interestingly upon Evelyn Fox Keller's discussion (1985) of "order" as a generalization of traditional notions of "law" in science. Latour (1988a) makes some interesting and perceptive observations and argu-


Now for the other difficult aspect of the mangle. We can return for a moment to the chicken debate. Collins and Yearley operate on a terrain that I find much more familiar than Callon and Latour's (one needs to read Callon and Latour's essay to appreciate this point fully), and here we run into the problem of the "posts." Collins and Yearley's arguments are familiar, I think, because they are modernist, humanist, dualist; Callon and Latour's are less familiar because they are postmodern (though Latour [1990, 1993b] does not like the term), posthuman, postdualist (if that is a word).39 To explain what is going on, I can dwell on the most perspicuous of these distinctions, humanist/posthumanist. Collins and Yearley's position, like SSK in general, is grounded in the traditional humanism of sociology as a discipline, inasmuch as it takes the human subject to be the center of the action. Whenever SSK detects a tendency for the action to be located elsewhere--a decentering of the human subject--it undertakes the police action already discussed. Any trace of nonhuman agency is immediately recuperated by translating it into an account of nonhuman agency that is attributed straight back to human subjects. Or, as we have seen, it is admitted that one can speak naively about nonhuman agency, but if one wants to do that, according to Collins and Yearley, one has to speak in a possibly even more familiar and antihumanist idiom, that of the scientists and engineers. Their discourse is a pure one of material agencies from which human agency is quite absent. (Thus Collins and Yearley's dualism.)

These humanist and antihumanist discourses run deeply through everyday thought, though they do not exhaust it. They are also the very stuff from which the traditional academic disciplines are created and that holds them apart (I come back to this in section 7.1). To be a traditional sociologist is to be a humanist; to be a physicist is to be an antihumanist (in technical practice, I mean). But the mangle, like the actor-network approach, corrodes the distinctions these discourses and disciplines enforce. It is not just that the mangle and the actor-network multiply sites of agency, trying to keep both human and nonhuman agency in view at the same time. Rather, in their somewhat different ways, the mangle and the actor-network insist on the constitutive


ments about explanation in the context of actor-network theory, though he couches them in the representational/semiotic idiom that I seek to avoid.

39. People usually say monist instead of postdualist; in the light of the discussions of cultural multiplicity in section 1.5, though, I think a better word would be "multiplist." My analysis is on the other side of two from one.


intertwining and reciprocal interdefinition of human and material agency. The performative idiom that I seek to develop thus subverts the black-and-white distinctions of humanism/antihumanism and moves into a posthumanist space, a space in which the human actors are still there but now inextricably entangled with the nonhuman, no longer at the center of the action and calling the shots. The world makes us in one and the same process as we make the world. This posthumanism is the second aspect of the mangle that thought tends to bounce off and even recoil from. Again, I make no apologies for it. It seems to me to be how things are in practice, an interesting but difficult observation that needs explicitly to be made, and the mangle helps to keep it in view.40 To speak for myself, I am not sure whether the temporal emergence of the mangle or its posthumanism is the harder to take. Perhaps it is the entanglement of the two, since my way of understanding the interconnection of human and material agency is essentially temporal.

In any event, to get clear on what is at stake I return to these themes of emergence and posthumanism as the book progresses. Reviewing the studies that follow, I try to be as clear as I can on the details of just how they exemplify temporal emergence and the decentering of the human subject in the extension of culture. From time to time, I also explicitly contrast my analysis of this process with traditional--humanist and nonemergent--accounts.41 My prime example of the latter is the approach I have already discussed most--SSK. And some explanation is needed for this expository tactic and choice of foil. I have tried, in general, to write this book constructively, laying out my own understanding of scientific practice as best I can. I have, however, come to the conclusion that the general trend of my arguments is best grasped by the kind of periodic confrontation with the more traditional views just men-


40. I finally grasped the significance of the humanist/posthumanist distinction in conversations with John Law and in reading his recent work (Law 1993, for example). Barbara Herrnstein Smith also helped me by pointing out a "residual humanism" in an earlier essay of mine (and the residue is arguably still there in the above discussion of human intentionality). Beyond the actor-network literature, I have found the works of Gilles Deleuze and Felix Guattari (especially Deleuze and Guattari 1987) and Michel Foucault very useful in thinking about the decentering of the human subject. Foucault (1972) offers a general elaboration of the themes of temporal emergence and the displacement of the human subject, though how he connects them is not clear to me. I link them via the preceding discussion of agency, a topic on which Foucault displays a principled reluctance to speak.

41. From now on I use "traditional" as shorthand for humanist, nonemergent, and representationalist approaches in science studies.


tioned. Among the latter, in most of these confrontations I single out SSK for three reasons. First, because I find it by far the most plausible and interesting nonemergent humanist account of science that is available (much more plausible and interesting than, say, accounts couched in terms of epistemic rule following--though I do discuss those, too, in section 6.4). Second, because I suspect that SSK has succeeded too well. I mentioned earlier the invaluable role that SSK has played in widening our conceptions of science as an object of study, but it seems to me that much of the work currently being done in science studies has remained stuck in the place where SSK left it in the early 1980s. Perhaps this is not the case for specialists in the sociology of science. Even the late r work of some its founders, like Steven Shapin, often bears little resemblance to SSK's prototypical form, which is why I tend to refer to "early" or "canonical" SSK in subsequent comparisons and confrontations. But many historians and philosophers who have become sensitized to science's social dimensions continue to treat SSK as if it were the last word on the subject (witness the current fashion for multidisciplinary eclecticism, discussed in section 7.1, or the dominance of interest explanations in STS). In taking SSK as a primary exemplar of nonemergent humanism, then, my aim is therapeutic: I would like to help end the sleep induced by this particular spell.42 My third reason for discussing SSK is a constructive reformulation of the second. SSK tends to treat the social as a nonemergent causal explanation of cultural extension in science. In contrast, my studies convince me that the social is bound up with science and itself subject to mangling in scientific practice. Unlike SSK, then, I am directly interested in processes of social transformation. I want to analyze them, and comparisons with SSK in particular can thus help me to foreground what is at stake in this aspect of my analysis.


There are angles on the mangle that remain to be mentioned, and the easiest way to get at them is to look ahead. The rest of the book is divided into two parts. Part 1, comprising chapters 2 to 5, revolves around


42. An alternative rhetorical strategy would be to describe the mangle (together with other nontraditional approaches to science studies) as a part of what SSK has become. Trevor Pinch, for example, has tried to persuade me of such a broad construal of SSK. Even following this line of thought, however, I think that it would remain important to make it explicit that SSK has taken an emergent and posthumanist turn away from its origins and canonical works. I also think that this construal tortures the words "sociology


four case studies intended to exemplify and elaborate upon aspects of the mangle. Part 2, chapters 6 and 7, follows up implications and leads in philosophy, social theory, and historiography of science.

As one would expect, chapter 2 is about the machinic base of science. It focuses on the construction of an instrument, the bubble chamber as a tool for elementary-particle physics. Following Peter Galison, I trace out the evolution of the bubble chamber from its first conception to its realization as a working device, and analyze its historical development as an exemplification of the mangle. This chapter thus puts some flesh on my ideas concerning the capture of material agency, modelling, and intentionality, the intertwined emergence of material and human agency in the dialectic of resistance and accommodation, and so on. As far as human agency is concerned, I talk first about the mangling of its intentional structure as discussed above. I then turn to a different aspect not so far discussed, emphasizing the mangling of the social contours of human agency--the transformation and interactive stabilization in practice of both the scale of social actors and their relations to other such actors. And I seek to clarify what is at stake in my analysis through a comparison between my account of human agency and traditional accounts framed in terms of enduring causes and constraints, emphasizing the emergent and posthuman displacements effected by the mangle. I can note here that appeals to notions of nonemergent constraint seem to be running rampant at present, even among authors otherwise opposed to traditional philosophy and social theory. I hope that my analysis can help stem the flood, and I continue the argument in chapter 5 (against the related notion of "limits") and in chapter 6 (in my discussion of relativism). Besides topics of agency, chapter 2 also begins my analysis of how knowledge can be understood within the performative idiom, arguing that representations of how the bubble chamber functioned were themselves mangled in practice.

Like chapter 2, chapter 3 is concerned with the machinic base of science and captures of agency, but there I am more directly interested in how knowledge is threaded into the machinic field of science--with how one can think representation in the performative idiom, with, espe-


of scientific knowledge." The move to the performative idiom strips knowledge of the special privilege that the K of "SSK" accords it; likewise the antidisciplinary tendency of the mangle (see above and section 7.1) argues against the disciplinary allegiance of the first S. The second S might stand, though I argue in chapter 5 and sections 7.3 and 7.4 that on my analysis science bleeds right into technology, the factory, and society.


cially, the production of scientific facts. The empirical study is of the construction of an instrument designed to investigate the existence of tiny particles called quarks, and of the use of that instrument to generate empirical knowledge of immediate theoretical relevance. An aspect of science that I mentioned earlier in this chapter but have since left undeveloped becomes crucial here, namely, its cultural multiplicity and heterogeneity. I argue that scientific practice is, in general, organized around the making (and breaking) of associations or alignments between multiple cultural elements, and that fact production, in particular, depends on making associations between the heterogeneous realms of machinic performance and representation, in a process that entails the emergent mangling and interactive stabilization of both. Articulated knowledge and machinic performances are reciprocally tuned to one another, I suggest, in a process that involves the artful framing of already captured material agency. At the end of the chapter, I draw upon other studies of science, first to elaborate an overall image of scientific knowledge as sustained in extended representational chains terminating in captures and framings of material agency, and then to argue for the mangling and interactive stabilization of the disciplined practices that envelop, sustain, and are sustained by such chains.

Chapter 4 continues the attempt to understand representation in the performative idiom, and is entirely about conceptual practice, the business of extending the conceptual apparatus of science. Its underlying concern is with how conceptual practice can have a shape: why is it not completely arbitrary or subjective? This question arises within my analysis because conceptual practice proceeds in the absence of any immediate engagement with the material world that could help constitute the kinds of dialectic of resistance and accommodation that the two preceding chapters dwell upon. So what, if anything, do scientists have to struggle with in the realm of concepts and thought? What lies between them and their goals? My answer to such questions depends upon seeing conceptual practice as again seeking to create alignments between multiple cultural elements, and locates the possibility of resistance in this process not now in material agency but in what I call disciplinary agency--the sedimented, socially sustained routines of human agency that accompany conceptual structures as well as machines. I show how disciplinary agency can play an analogous role in conceptual practice to that of material agency in material practice. And I show, too, how disciplinary agency is itself mangled in conceptual practice. I should note that my example of conceptual practice comes from mathematics


rather than science proper: it concerns Sir William Rowan Hamilton's formulation of the "quaternion" system in 1843. I am happy that my analysis of the mangle can be extended to mathematics, but I hope that this chapter will also be read as an indication of how conceptual practice in science, including the realm of scientific theory, might likewise be conceived. The chapter ends with a summary of the analysis developed in chapters 2 to 4, followed by a postscript in which I discuss David Bloor's SSK-style analysis of Hamilton's work. Bloor suggests that Hamilton's metaphysical orientation was dependent upon his social position and interests; my suggestion instead is that Hamilton's metaphysics was emergently and posthumanly mangled in his technical practice.

Chapter 5 moves us into a rather different space, so before I come to it I want to make a general comment on chapters 2 to 4. These chapters constitute the empirical body of my performative understanding of science. They run upward from the construction and use of machines and instruments, through the production of facts and their engagement with theory, to the quasi- autonomous dynamics of thought and representation. That sounds like quite a lot, but, having worked through and reflected upon these studies I am conscious of what a sparse set of examples they are in relation to what I have come to understand as the immense complexity of scientific culture and practice. More particularly, it is clear that my terms of engagement with practice are tuned to the physical/mathematical sciences (and probably many branches of engineering) rather than to other sciences. The word "machine" comes less and less naturally as one moves from physics even to chemistry, let alone to fields like biology or the social sciences. I confess, though, that I am not very disturbed by this. As the writings of the actor-network amply demonstrate, ideas about nonhuman agency are more readily graspable in respect of the biological than the physical sciences, and the work of Michel Foucault points the way to a parallel analysis of the human sciences. One can see well enough, then, how my account of scientific practice might be extended beyond its present domain of exemplification (which is not to say that detailed explorations would not be illuminating).43


43. For example, I find many suggestive connections between my analysis of the mangle and Foucault's discussion (1979) of the prison. In my terms, the prison can be seen (like other institutions) as a site for capturing, channeling, and framing the agency of a particular segment of the human population, and much of Foucault's analysis relates to a kind of institutional tuning or mangling, in which the social sciences have themselves emerged and continue to be mangled. More generally, I think that my discussion of the


Now for chapter 5. This breaks the sequence established by chapters 2 to 4. It is not about science per se. It is about technology and the workplace, about industry, production, capital, and labor relations. It is, to be specific, about the introduction of the new technology of numerically controlled machine tools into the factory. My principal reason for its inclusion is that it shows that my analysis of the mangle can be put to work outside the realm of science. As in chapters 2 and 3, a notion of capturing material agency is at the heart of chapter 5, as is that of the mangling of the social--here meaning the formalized and disciplined roles and relations of workers and management. The point that I want to emphasize in this chapter is, then, that a performative analysis like the mangle runs very smoothly and directly between the worlds of science, technology, and work: it promises an integrated vision of STS. A single example cannot, of course, make the case, but the overall conclusion is something that the actor-network has already reached (see, for example, Latour 1987), and I want to show how we can arrive at it in the performative idiom. A subsidiary objective of this chapter is to continue the confrontation with traditional understandings of practice. I take my empirical information from David Noble's Forces of Production, a book in which Noble offers a quite different interpretation from mine of the events at issue. His is SSK-like, a traditional, nonemergent, and humanist interpretation based-upon enduring social actors and their interests, in which a notion of enduring limits to practice plays a crucial role; mine is emergent and posthumanist, foregrounding the transformations in the social that are part of the encounter with material agency. A critical comparison of our two approaches serves, therefore, to highlight what is more generally at stake in the move to the mangle.

The two chapters that make up part 2 of the book take those of part 1 as their point of departure. Chapter 6 is especially concerned with appreciations of scientific knowledge that follow from the mangle. It begins by trying to get to grips with the problematic of realism, which I take to center on the nature of the purchase our knowledge has on the world and vice versa. I argue that the mangle offers us a realistic appreciation of scientific knowledge inasmuch as it demonstrates the nontrivi-


mangling of representational chains in chapter 3 could readily be generalized to the relays and alignments in chains of power/knowledge that Foucault has thematized. In this connection, see the Foucauldian analyses of developments in the social sciences, management and accountancy by Hopwood (1987), Miller (1992), Miller and O'Leary (1994), Rose (1990), and Rose and Miller (1992).


ality of the construction of representational chains terminating in captures and framings of material agency. I call this the pragmatic realism of the mangle, to distinguish it from the correspondence realism that defines the traditional realist-antirealist debate in philosophy. I argue that pragmatic realism is not, in the first instance, a position in the debate over whether representations correspond to nature, but that it subverts that debate. Once one appreciates the pragmatic realist engagement of knowledge with the world, questions of correspondence no longer seem pressing. One can, however, make trouble for correspondence realism on the basis of its pragmatic namesake. I argue that it is consistent with pragmatic realism to suppose that the world will support an indefinitely diverse set of ontologies and bodies of knowledge, each terminating in its own particular field of machines, and I offer historical exemplification of this possibility. At stake here, then, is an idea of machinic incommensurability that cannot be formulated in the representational idiom but that makes it hard to suppose that any particular scientific ontology exists in a special relation of correspondence to nature itself.

I return to the discussion of incommensurability in chapter 7, but the rest of chapter 6 addresses issues concerning objectivism about scientific knowledge and alternatives to it. I begin by defining the sense in which the mangle is an analysis of the objectivity of scientific knowledge. My argument is that maneuvers in the field of material and disciplinary agency immediately delocalize and confer a degree of objectivity on the products of scientific practice even at the individual level. On my analysis of practice, it is far from the case in science that "anything goes." However, as one would expect from my comments on incommensurability, the objectivity of the mangle is itself consistent with a certain relativity of scientific knowledge, a dependence of knowledge upon the conditions of its making, evaluation, and use. Scientific knowledge, on my account, is both objective and relative. Care needs to be taken, however, with these terms. I contrast my analysis of objectivity with the traditional philosophical one that looks to nonemergent and humanist rules of reason to guarantee the objectivity of theory choice. I argue that the mangle subverts this picture, and that whatever rules one is inclined to see as operative in science have themselves to be seen as in the plane of practice, and as liable to emergent mangling in one and the same process as all of the other heterogeneous elements of scientific culture. I then make a similar argument concerning traditional accounts of the relativity of science in terms of nonemergent interests or constraints (or limits)


that have been held to specify the dependence of knowledge upon the social. The thrust of the studies in part 1 is that whatever characteristics of the social one cares to identify need, again, to be seen as subject to mangling, as themselves remade in practice. On my account, then, there is no enduring substantive link to hang onto in grasping the relation between what culture is and what it is to become--neither epistemic rules nor properties of the social furfill this function. There is only the mangle, and what it cares emergently to grind out. And this, in turn, speaks of an irremediable historicity of scientific knowledge (and culture in general): what counts as knowledge now is a function of the specific historical trajectory that practice has traced out in the past. To emphasize this point, the chapter concludes by continuing the discussion of the quark-search experiments of chapter 3 into a controversy that came briefly to surround them. Controversy is a classic site of articulation for SSK, which argues that divergent accounts of nature call for distinctively sociological explanation. In contrast, I suggest that nothing is needed to explain the divergence at issue beyond the mangle and the real-time contingencies of its working. The conclusion of this phase of the discussion is, then, that scientific knowledge is objective, relative, and historical, all at once.

Chapter 7 turns from the study of scientific practice to the practice of science studies, and addresses two main topics. One concerns possibilities for synthesis among the fields that make up science studies. I note that traditional explanatory structures lead directly into the well-known divergences and battles between disciplinary (philosophical, sociological, and so on) accounts of science and that, in their own terms, these divergences can only be overcome by adding up competing accounts in a strategy that I call multidisciplinary eclecticism. The mangle can instead contribute to, and offer a rationale for, a new and genuine antidisciplinary synthesis--deflating traditional schemes by displacing disciplinary variables (rules, interests, and so on) onto the same plane as technical scientific practice itself. I then turn to what must be a central component of the new synthesis: cultural studies of science, as defined in section 1.1. I argue that cultural studies and the mangle should be seen as orthogonal but complementary approaches to science studies, each contributing to a common understanding of science along its own axis; and I try to show how the new synthesis in science studies thus delineated opens out into a broader antidisciplinary synthesis of which the field of cultural studies per se is a central marker.

The second principal topic of chapter 7 is the question of how to get


from the micro to the macro. Noting that the studies of part 1 relate to the practice of individuals or relatively small groups, I enquire here into the implications of the mangle for how we might think about practice at a more aggregated social level. Having had my say on topics in philosophy and social theory earlier in the book, at this stage I concentrate on historiography, on how, especially, we might write history in the large. I suggest that the move from the representational to the performative idiom points toward an integrated historiography of STS in which, to put it iconically, the Industrial Revolution takes over the special place occupied by the Scientific Revolution in traditional history of science. I then exemplify the possibility of getting to grips with macrohistory as macromangling in a discussion of the intersection of scientific and military enterprise in World War II. My argument here is that my analysis is scale invariant: that practice at any level of aggregation can be understood in terms of emergent and posthuman mangling. I further note that the study of macromangling promotes a desirable kind of double vision, a historiographic sensitivity both to the mangle-ish character of history in general and, at the same time, to the specificity of particular historical developments and formations.

Chapter 7 concludes with two postscripts--optional extras. In the body of the book, I take for granted certain standard but nevertheless circumscribed conceptions of what people and machines can do. In the first postscript, I note that outside industrialized societies quite nonstandard ideas of human and material agency are common, and I argue, first, that my analysis could readily be extended to nonstandard agency and, second, that a recognition of nonstandard agency points to a very radical form of incommensurability, an incommensurability of powers. The second postscript develops an idea of the mangle as a Theory of Everything. If one deletes the specifically human features of my analysis--in particular, the discussion of human intentionality broached earlier--I cannot see why it should not apply to the world itself: to chemical, biological, geological, cosmological systems, and whatever. The stumbling block is, of course, the rival nonemergent accounts offered by the existing sciences, and the book ends with some thoughts on how to get around them.