u n i v e r s i t y o f c o p e n h a g e n Beauty and Bacteria Visualizations in Molecular Ecology Sommerlund, Julie Published in: Configurations Publication date: 2007 Citation for published version (APA): Sommerlund, J. (2007). Beauty and Bacteria: Visualizations in Molecular Ecology. Configurations, 12(3), 375 - 400. Download date: 06. apr.. 2021 https://curis.ku.dk/portal/da/publications/beauty-and-bacteria(a234f792-8985-4c91-9662-59eb5aa01bbb).html Beauty and Bacteria: Visualizations in Molecular Microbial Ecology Sommerlund, Julie. Configurations, Volume 12, Number 3, Fall 2004, pp. 375-400 (Article) Published by The Johns Hopkins University Press DOI: 10.1353/con.2007.0004 For additional information about this article Access Provided by New Copenhagen University Library at 10/10/12 12:44PM GMT http://muse.jhu.edu/journals/con/summary/v012/12.3sommerlund.html http://muse.jhu.edu/journals/con/summary/v012/12.3sommerlund.html 1. Biofilms are communities of bacteria—often comprising different species—that lump together on a surface and protect themselves with a layer of slime. Biofilms can be found anywhere—in water pipes, on teeth, in refrigerators—but only recently did researchers start creating and studying them in laboratories. Previously, scientists worked primarily with one species of bacterium at a time, which they suspended in wa- ter. Today there is a growing awareness that this is not how bacteria exist in nature— where they most commonly live in biofilms. For further discussion of the challenges that arise from biofilm research, which is by definition transdisciplinary, combining ecological methods with techniques from molecular biology, see Julie Sommerlund, “Multiple Classifications and Research Practices: Classifications in Molecular Microbial Ecology,” Social Studies of Science (forthcoming). Introduction In November 2000, Nature, the prestigious scientific periodical, ran a news feature entitled “Slimebusters.” The feature dealt with biofilm, and was illustrated by a picture of biofilm made by a group of microbiologists working with Confocal Scanning Laser Micro- scopes (CSLM) at the Danish Technical University (Fig. 1).1 Having a picture published in Nature was somewhat of an event for these re- searchers. At the time, I was doing observational studies in the group, and the researchers told me that this would probably be the only time that anything they produced would be published in Na- ture. They joked about it–that their only publication in Nature was a picture illustrating an article called “Slimebusters,” and not a “real” scientific article. I did not understand what was “not really scien- tific” about it. I asked the professor in the group what the problem was, and he replied that the pictures they made had become “too 375 Configurations, 2004, 12: 375–400 © 2007 by The Johns Hopkins University Press and the Society for Literature, Science, and the Arts. Beauty and Bacteria : Visualizations in Molecular Microbial Ecology Julie Sommerlund The Danish School of Design 376 Configurations artistic”; the group had even been approached by an experimental jazz-saxophonist, who wanted to use the pictures as a visual back- ground at a concert. Not that it matters as such, the professor said, but the group is becoming famous for the aesthetic qualities of its pictures, not for its contribution to “hard science.” Interesting di- chotomy, I thought. In this paper, I intend to pursue the assumed dichotomy between science and aesthetics. One of the most remarkable features of the boundary between the two is that these researchers seem to disre- gard it in practice and to mix science and aesthetics in a quite radi- cal way, while their verbal reflections of their practice seldom recog- nize this. In other words, there seems to be a major discrepancy between what the researchers say and what they do. In the follow- ing, I will trace the discourses and practices of science and aesthetics by analyzing the making of the group’s emblematic photographs, the CSLM-images. In the first part of the paper, I will examine how bacteria in biofilms are modified in a fashion that makes them visi- ble; in the second part, I will analyze how the pictures are edited and made understandable. Science, Art, and Aesthetics Like the stated aim of Alberto Cambrosio, Daniel Jacobi, and Pe- ter Keating, my aim with this paper is to “contribute to the analysis of “nonart” (in the present case scientific) images.”2 The critical dif- ference between art and nonart is arguably that aesthetics is an ex- plicit actor in the construction of art, whereas aesthetics plays only an implicit, or even hidden, role in the construction of science. In the following, I will discuss the relations between these three phe- nomena: science, art, and aesthetics. Traditionally, science is perceived as focusing solely on content while being completely oblivious to form, whereas aesthetics is per- ceived as concentrating exclusively on form and disregarding con- tent. This division is so crude that it borders on caricature, and many scholars have written about the complexity of the relation between content and form.3 Here, however, I will stay with the more “tradi- 2. Alberto Cambrosio, Daniel Jacobi, and Peter Keating, “Arguing with Images: Paul- ing’s Theory of Antibody Formation,” Representations 89 (2005): 94–-130, on p. 96. 3. Prominent among these is Hayden White, The Content of the Form (Baltimore/Lon- don: Johns Hopkins University Press, 1987). The form analyzed is the narrative, and not visual imagery as in this paper; however, the point in this case—that content and form cannot be regarded as discrete parts, but rather are intimately intertwined—goes for the present study as well. tional” stance on content and form for a while, because this is the stance that I have met in the laboratory when the researchers com- mented on their work, and therefore it will have a profound impact on the analyses that follow. Important consequences of the traditional view of content and form—that they are different, and that aesthetic concerns have no place in science—are the divergent ways of regarding the constructive processes leading to the end product, be it scientific or aesthetic: Aesthetics generally highlights the processes leading to the final product. Skill and technique are often major elements of the work, and are experienced as pleasurable to the viewer/listener/receiver. Thus, appreciating the violinist’s ability to move her fingers rapidly and precisely is as vital a part of the experience when listening to Pa- ganini as is the music “in itself”. Appreciation of the painter’s ability to apply paint to canvas and his ingenuity in coming up with new ways of doing so is as important a part of the experience when look- ing at a Van Gogh painting as are the colors and shapes “in them- selves.” Knowing that Bach composed Das wohltemperierte Klavier Sommerlund / Beauty and Bacteria 377 Figure 1. Christensen, Haagensen, Heydorn & Molin, Biofilm consisting of Acinetobacter and P. putida (Courtesy of Applied and Environmental Microbiology and the authors, re- published from “Metabolic Commensalism and Competition in a Two-Species Microbial Consortium,” May 2002, Vol. 69, no. 5. p. 2496–202) around the twenty-four scales, presenting a prelude and a fugue in each, adds grandeur and overarching structure to the music “in itself.” In science, however, the underlying processes behind the inven- tion of scientific facts and data presented in publications are rou- tinely suppressed. Working hypotheses are left unmentioned, failed experiments likewise, and—as we shall see—data are presented in styles that suppress the author as an individual and replace him or her by an impersonal, all-knowing author as anchor of the text.4 Thus, in science, the constructive processes deduct validity and weight from the end product; in aesthetics, the constructive processes add to the experience of the end product. The picture in Nature points directly to this dichotomy: If science is investigation of content, and aesthetics is invention of form, then the constructive aspects of the pictures’ aesthetic qualities challenge their scientific status. The issue of science and aesthetics is central to a growing de- bate within Science Studies that focuses on visualizations in science. Examples are the anthologies Representation in Scientific Practice5 and Picturing Science, Producing Art,6 as well as a number of recent articles. Among the writers participating in this debate, Bruno Latour is probably one of the most influential, in regard to both the scientific community in general and this paper in particular. Latour has writ- ten extensively about science,7 and has also touched upon the topics of aesthetics, construction processes, and inscription devices,8 which are all important to this paper. In an article entitled “How to be Iconophilic in Science, Art and Religion,” he comments directly on the science/aesthetics debate.9 He claims that matters of construc- tion lie at the very core of art, which is in agreement with the points I made above. Art history, which explains how specific works of art 378 Configurations 4. It is a common point within Science Studies that the sciences become sciences, aloof from other epistemological enterprises, by erasing the traces of work and contin- gency that have led to the data presented. See, e.g., Bruno Latour, “How to Be Iconophilic in Science, Art and Religion,” in Picturing Science, Producing Art, ed. Caro- line A. Jones and Peter Galison, (London.New York: Routledge, 1998), pp. 418–441. 5. Michael Lynch and Steve Woolgar, eds., Representation in Scientific Practice (Cam- bridge, Mass.: MIT Press, 1990). 6. Jones and Galison, Picturing Science, Producing Art (above, n. 4). 7. See, e.g., Bruno Latour and Steve Woolgar, Laboratory Life (Princeton: Princeton Uni- versity Press, 1979); Bruno Latour, Science in Action (Cambridge, Mass.: Harvard Uni- versity Press, 1987); idem, Pandora’s Hope (Cambridge, Mass./London: Harvard Univer- sity Press, 1999). 8. Inscription devices are technologies that translate phenomena into signs on paper. 9. Latour, “How to Be Iconophilic” (above, n. 4). were constructed, is not perceived as challenging the works’ status as art; rather, it adds to the pleasure of art: the more you know about the making of a certain painting or movie, the more you will enjoy it. Likewise, claiming that there is no work done in creating art would provoke most artists, who state again and again that their work is ex- actly that: work.10 This is not the case in science, which is not per- ceived as constructing facts, but as uncovering or discovering facts. Thus, explicating the work done by inscription devices will be per- ceived as challenging to the sciences, for it will point to the purification work done by the inscription devices—which again indicates the com- plex and contingent layers of work that make inscriptions possible.11 In Science in Action, Latour describes how the world of the labora- tory opens up to the spectator interested in what lies “behind” the inscriptions when confronted with controversy: “What is behind a scientific text? Inscriptions. How are these inscriptions obtained? By setting up instruments. This other world just beneath the text is in- visible as long as there is no controversy.”12 In the Molecular Micro- bial Ecology Group (MMEG), I seemed to have stumbled upon one such controversy, opening up to investigations of the “world be- neath”—the laboratory and its instruments. Cambrosio, Keating, Jacobi, Lorraine Daston, and various others have written a substantial body of texts analyzing and discussing im- agery in science.13 These texts are an important backdrop and inspi- ration to this paper. Here, I will single out two that have had specific influence on my analyses—but that does not mean that the remain- ing articles have been forgotten. The article “Of Lymphocytes and Pixels: The Techno-Visual Pro- duction of Cell Populations”14 deals with visualizations created by Sommerlund / Beauty and Bacteria 379 10. For further discussion, see also Bruno Latour, “The Promises of Constructivism,” in Chasing Technoscience: Matrix for Materiality, ed. Don Ihde (Bloomington: Indiana Uni- versity Press, 2003), pp. 27–46. 11. Latour, Science in Action (above, n. 7), p. 64 12. Ibid, p. 69. 13. See, e.g., Peter Keating, Alberto Cambrosio, and Michael MacKenzie, “The Tools of Discipline: Standards, Models, and Measures in the Affinity/Avidity Controversies in Immunology,” in The Right Tools for the Job, ed. Adele Clark and Joan Fujimura (Prince- ton: Princeton University Press, 1992), pp. 312–355; Peter Keating and Alberto Cam- brosio, “`Going Monoclonal’: Art, Science, and Magic in the Day-to-Day Use of Hy- bridoma Technology,” Social Problems 35 (1988): 244–260; Peter Keating and Alberto Cambrosio, “Biomedical Platforms,” Configurations 8 (2000): 337–387. 14. Alberto Cambrosio and Peter Keating, “Of Lymphocytes and Pixels: The Techno- Visual Production of Cell Populations,” Studies in History and Philosophy of Biological and Biomedical Sciences 312 (2000): 233–270. means of a machine called a FACS (Fluorescence-Activated Cell Sorter).15 The FACS distinguishes different species of bacteria “me- chanically”: the bacteria fluoresce in different colors, and the FACS re- sponds to the differences in type of fluorescence. This differentiation is performed “manually” in a conventional fluorescence microscope such as the CSLM. In spite of this difference, the central point of Cambrosio and Keating’s article—that imaging via the FACS tech- nique breaks down the difference between representation and inter- vention—also goes for the CSLM technique discussed here. Accord- ing to Cambrosio and Keating, “imaging as embodied in the flow cytometric practices does not simply break down Walter Benjamin’s distinction between manual (for example, drawings and engraved diagrams) and mechanical (for example, photographs and micro- graphs) reproductions, but goes beyond that to collapse the epistemic distinction between intervening and representing.”16 This point is very similar to the one I will present in this paper, for the collapsing of “intervening and representing” can be seen as related to the implo- sion of reality and representation that is implied in the transforma- tion of nonoptical events into optical ones. Thus, the point sug- gested by Cambrosio and Keating would be common to fields as different as astronomy, high-energy physics, and microbiology. Another paper by Cambrosio, Jacobi, and Keating focuses on Paul Ehrlich’s images of antibodies and points out that these images rep- resented entities on which there was no consensus: “What made the use of visual imagery so controversial, was that it defined the rele- vant entities by the very process of representing them.”17 There are many similarities between this situation and the situation my re- searchers find themselves in, for biofilm is a new object of research with new methodological tools offered by molecular biology to eco- logical studies. Thus my paper and Cambrosio, Jacobi, and Keating’s 380 Configurations 15. This technology is central in the MMEG laboratory as well, although it was ac- quired only when I was finishing my work there; consequently the FACS is not in- cluded in my analyses. 16. Cambrosio and Keating, “Lymphocytes and Pixels” (above, n. 14), p. 235. Walter Benjamin was an influential German philosopher who published the ground-breaking article “The Work of Art in the Age of Mechanical Reproduction” in 1936, reprinted in Illuminations (London: Random House, 1999), pp. 211–244. The article discusses the technologies of film and photography, and how these technologies influence the sta- tus of the work of art and its special, authentic “aura,” which is constituted by the di- rect physical link between the work and the artist. 17. Alberto Cambrosio, Daniel Jacobi, and Peter Keating, “Ehrlich’s ‘Beautiful Pictures’ and the Controversial Beginnings of Immunological Imagery,” Isis 84 (1993): 662–699, on p. 666. share the focus on how representation shapes the entities repre- sented. The difference is mainly that theirs is a historical study, while mine is real-time and material. Upon this note, I will return to the Molecular Microbial Ecology Group. The Story of Acinetobacter and Pseudomonas putida After talking to the professor about the picture from Nature, I went on to ask the young researcher who had made the picture what was so special about it. He told me that it was actually a part of a se- ries of six, and that it was the story narrated by these six pictures rather than the single picture that was interesting. I asked him about the story, and he gave the following account: (Figure 2) The series consists of six pictures (Fig. 2) that represent particular phases of development of biofilms consisting of Acinetobacter (red) and Pseudomonas putida (blue). The researcher and his colleagues had a hypothesis about these species: that together they were able to de- grade the organic solvent toluene. When the two species grow on a bases of benzyl alcohol (replacing toluene), they go through differ- ent stages of cooperation: Acinetobacter is able to metabolize benzyl alcohol and excrete benzoate, which P. putida can metabolize. From the outset, the two bacteria live almost evenly scattered over the sur- face (A). However, it soon becomes obvious that P. putida is attracted to Acinetobacter, probably because of the excreted benzoate—this is visible because the active cells of P. putida (those that eat) shine with a greenish light. The next day (B), Acinetobacter starts to develop mi- crocolonies, and P. putida gathers around these colonies. On the third day, large microcolonies of Acinetobacter are formed, with active P. putida cells around and on top of them (C–D). On the sixth day of the development of the biofilm, the large microcolonies have disinte- grated, and Acinetobacter starts to migrate into the higher levels of the substratum (E–F). This probably happens because P. putida is so close to the Acinetobacter that it no longer gets the nutrients it needs. At first, I was simply baffled by the story, as I had no idea how the single picture-frames A–F were connected to the stages in the story. The exploration of these connections is very much what my coming analyses will consist of. Second, I was also a bit disappointed with the story in itself—I had expected something more dramatic after the introduction (it is not the single picture, it is the story that is interesting), perhaps a classic story of the rise and fall of a community.18 I had hoped that Sommerlund / Beauty and Bacteria 381 18. However undramatic, the story does have the basic characteristics of a narrative: it is structured in time; it has a fixed beginning, middle, and end; and it has a specific 382 Configurations Figure 2. Haagensen, The Story of Acinetobacter and P. putida (unpublished images: Cour- tesy of the author). the researcher would tell me more about the aesthetics/science schism. But to the young researcher the term “story” seemed to be more prosaic than it was to me; or perhaps the story of Acinetobacter and P. putida contained high drama within the context of microbiol- ogy, and I was simply not able to understand it. Later, I was to discover that even for me, there was high drama to be found in this story: the translation of the story to a single picture in many ways serves as an analogy to the process of translating and purifying scientific practical work into Science. Thus, the compli- cated relation between image and narrative implied by the re- searcher had an impact on many topics other than that of Acineot- bacter and Pseudomonas. More about that later on. The Construction of a Sign System I went on to ask the researcher who had made the picture if I could spend some days in the laboratory with him to see how the biofilm pictures were made. In the laboratory, it soon became appar- ent that a major issue when working with biofilm is to make visible the things you want to demonstrate. First, the researchers want the different species of bacteria to be distinguishable. Even when the bacteria grow species by species in a petri dish, it can be difficult to tell them apart: some may be oblong and others circular, and that is about all the difference there is. When more species are grown to- gether in a glass chamber—which is how biofilms are grown in the laboratory—you have no chance of telling them apart by morpho- logical differences. Second, the researchers want to be able to see whether bacteria of different species are living next to each other or to- gether. For instance, the biofilm depicted in Nature is interesting be- cause the two species may feed off each other. How do you show that? The young researcher needed a representative system: a vocab- ulary to tell the story of the biofilm.19 The Molecular Microbial Ecology Group laboratory has made the construction of sign systems somewhat of a specialty: The re- searchers use probes to mark the different strands of bacteria with Sommerlund / Beauty and Bacteria 383 causality. It should be noticed, though, that the causality originates from the bacteria rather than the “author”—the researcher. In this respect, reality and the representation of it are intermingled in the most intimate way. 19. The most commonly discussed sign system is human language. Umberto Eco, in A Theory of Semiotics (Bloomington: Indiana University Press, 1976), points to many more, such as zoology, olfactory signs, tactile communication, medicine, music, plot structures, mass communication, and written and visual languages. In The Fashion Sys- tem (1969; London: Cape, 1985), Roland Barthes also mentions the fashion system, and I will add scientific systems to the list. colors. These probes consist of specific DNA sequences that hook on to bacterial RNA, and they are marked with different fluorocromes— that is, with different colors. The use of probes is routine in the lab- oratory, and the probes are commercially available. How the com- mercial laboratories make the probes and how the probes actually connect to the bacteria remain a black box to the researchers20—they simply order the probes from a catalogue, as others would order art supplies.21 By means of these probes we enter the world of significa- tion, the making of meaning—a world that can fruitfully be dis- cussed using concepts from semiotics. Therefore, I will turn to semi- otics and discuss the color marking of bacteria in this light. A basic claim within semiotics is that the signified (the concept of Acinetobacter) is arbitrarily connected to the signifier (that which refers to the concept, in this case the word “acinetobacter,” or, in a picture, a red dot).22 This means that another sound or different let- ters would be just as efficient in referring to the bacteria as “acineto- bacter,” as would a different color than red. The connection is not given by nature, but is the result of a cultural and arbitrary defini- tion. The sign system considered here—the color-marking of strands of bacteria, and the visualization of their level of activity—is special, for the link between signifier and signified has not yet been estab- lished. The unmodified, wild-type bacteria are out of reach of the re- searchers and only come within reach through the process of signi- fication, a process that can be followed in the laboratory and that leaves no doubt as to the arbitrariness of the sign. This is an example of how turning nonoptical events into optical events also means re- moving boundaries between representation and intervention. 384 Configurations 20. For further discussion of scientific techniques and their becoming “black boxes,” see Kathleen Jordan and Michael Lynch, “The Sociology of a Genetic Engineering Tech- nique: Ritual and Rationality in the Performance of the ‘Plasmid Prep,’” in Clark and Fujimura, Right Tools for the Job (above, n. 13, pp. 77–115. The article is an analysis of the “plasmid prep,” a technique that is also used extensively in the laboratory in ques- tion, and that has a status in many ways equal to the ones I discuss here. 21. In Laboratory Life (above, n. 7), Latour and Woolgar use Bachelard’s concept of “rei- fied theory” to designate this type of technology. 22. Ferdinand de Saussure perceived the sign as a whole comprising a signifié (concep- tual content; henceforth, “signified”) and a signifiant (expression; henceforth, “signi- fier”). The relation between signified and signifier is arbitrary, which means that there is no natural connection between these two sign levels. By denaturalizing the connection between signified and signifier it is possible to discuss the human language—as well as other sign systems—as classificatory systems. Poststructuralists have later claimed that not only is the relation between signified and signifier arbitrary, the signified itself is arbitrary and is constituted by the signifier—not the other way around. See Ferdinand de Saussure, Course in General Linguistics (1916; London: Duckworth, 1992). In the laboratory, the connection between signified (“bacteria”) and signifier (the colored dot) was made as follows: The researcher set up the flow chambers containing the biofilm in a system of tubes and pumps that ensured the circulation of nutrients through the sys- tem. The biofilm was then left to grow for a few days. When it was ready to be photographed, the researcher turned off the pump and removed the flow chambers containing the biofilms that were to be photographed; he then took a syringe filled with ice-cold fixative (silicone rubber) and injected it into the flow chambers, a procedure that killed the bacteria in the biofilm. The six pictures represent stages in the development of a biofilm. But making a CSLM-picture kills the biofilm, and hence the single photographs in the series are of different biofilms. Thus, the picture- series representing the development of one microcolony in one biofilm is made using six different biofilms. The researcher then applied the probes to the silicone with an- other syringe and told me that they would spread through the sili- cone and hook onto the RNA of the bacteria, thus making the or- ganisms in the biofilm visible. In the case of Acinetobacter and P. putida, two probes were added to the silicone (red for Acinetobacter, blue for P. putida), and this way the two actors of the story were put in place. However, in order to tell the story of the two strands of bacteria, the researcher needed yet another color code. The actors had been coded red and blue, but an indicator of interaction was necessary to combine the two codes into a story line. This procedure had been performed before I started my observations: the interaction indicator had been spliced into the DNA of the two strands of bacteria con- tained in the biofilm; consequently, the following is based on what the researcher told me, and not on my own observations.23 The in- dicator is called “gfp” (green fluorescent protein), and it is as institu- tionalized as the probes mentioned above.24 Gfp was originally iso- lated from a fluorescent jellyfish, Aequorea vintoria; it is used for gene encoding, or more specifically to show when a gene is induced, or “turned on.” In the case of Acinetobacter and P. putida, gfp can be used to show whether P. putida’s metabolic system is more active when placed close to Acinetobacter than it is elsewhere. In other Sommerlund / Beauty and Bacteria 385 23. Even if I did not observe the researcher splice gfp into Acinetobacter or P. putida, I feel fairly familiar with the procedure, as this was one of the experiments I performed myself when being “trained” prior to making observations. 24. During the past five to ten years, the natural gfp protein has been heavily mutated; thus, bfp (blue fluorescent protein), yfp (yellow fluorescent protein), etc. are all com- mercially available—again, just like art supplies. words, if P. putida can eat what comes out of Acinetobacter, they will shine with a greenish light. The practice of color encoding is routine and black-boxed, but every time a researcher performs the procedure it is at least theoreti- cally possible to create a new color code and make the invisible visible in a different way. There is no “natural” line between visible and in- visible; rather, the boundaries are based on the practical choices that researchers make in specific cases. Even more strikingly, though the goal of this experiment may be to examine the toluene-degrading capabilities of microbial communities, laboratory time is primarily spent working with signification. There is no mystery to this, for there is no way of examining toluene-degrading capabilities without in some way making them visible or detectable. In the words of Hans-Jörg Rheinberger, “Representation is ‘eventuation’ (it is about intervention, invention, and the creation of events).”25 This questions the practical relevance of the schism between aes- thetics and science: in practice, there is no clear line separating the two. However, the researchers’ verbalizations continue to differenti- ate sharply between them. This differentiation will be the object of discussion below. “Subjectivity” When the researcher chose red and blue and not two other colors to mark his bacteria, this was simply because “these colors stand out against each other nicely”—as he said in the laboratory. Later, in an interview, he developed the point further: D: For Acinetobacter, I chose a red probe, but I might as well have chosen a green one or a blue one. . . . JS: Then why did you choose red? D: I chose the red probe because that is what people have done earlier. And people have used the blue for putida. JS: So, it’s becoming a code? D: It is becoming a color code—or at least it is within this group. I do not know if other groups use other codes. It is also a question of what you have in the freezer. And of course, you should not use green because the gfp monitor is green, so of course you will have to think about that. But otherwise, it is about what is in the freezer, and what other people have done with the bacteria you are working with. 386 Configurations 25. Hans-Jörg Rheinberger, Toward a History of Epistemic Things—Synthesizing Proteins in the Test Tube (Stanford: Stanford University Press, 1997), p. 108. What struck me—both during my observations in the laboratory, and later while doing the interview—was that the researcher seemed to dismiss all questions about color and choice of motif. They seemed to be unimportant to him, and he could not understand why I found the subject interesting.26 Nevertheless, this specific re- searcher is known within the group to be a specialist when it comes to constructing and representing biofilms. To a layperson, the pic- tures are intriguing because of their beauty. I was mystified that the young researcher did not show more pride in this aspect of the pic- tures, and in the fact that even outsiders found them interesting.27 One way of understanding this could be to consider what the re- searchers refer to as the subjective quality of the photos. Subjectivity, when defined by these researchers, seems primarily to be about ex- plicitness in choice of motif. The choice of photographing one micro- colony rather than another is perceived as being highly subjective. In the following, another researcher from the group explains why this is the case: Because it is really very chaotic what goes on in one of those biofilms. So, it is very . . . I don’t know if the others have told you, but often when they make pictures for publications, they scan more biofilms to find the best place that shows exactly what they want to show. Especially . . . acinetobacter and putida . . . those nice pictures they have. Everybody who works here knows that these kinds of pictures that make it to that kind of publication . . . they have spent maybe half an hour scanning the entire biofilm to find the best place, and then you say, it looks like that and that, and you’ve got this gorgeous picture to underline your hypothesis. But somehow it is not very scientific, doing that, but everybody does it. I guess somehow it’s OK. The researcher quoted above regards “subjectivity” as being in op- position to “science.” Almost all the researchers of the group echo Sommerlund / Beauty and Bacteria 387 26. Alex Pang notes a similar inclination toward aesthetic considerations in astron- omy, and quotes a late nineteenth-century astronomer as saying, on the introduction of photography to astronomy, that “stars should henceforth register themselves”; Pang goes on to state that “aesthetics would not even exist in this new order. `Their pictor- ial beauty,’ the Review [“Astronomical Photography,” Edinburgh Review, 1888] said of astronomical photographs, ‘is the least of their merits,’ and comparisons of pho- tographs would be made without considering whether one was more beautiful or strik- ing than another” (Alex Pang, “Technology, Aesthetics, and the Development of As- trophotography at the Link Observatory,” in Inscribing Science, ed. Timothy Lenoir [Stanford: Stanford University Press, 1998], pp. 223–249, on p. 224). 27. Not all researchers are as shy of speaking of the aesthetic qualities of their work. For instance, James D. Watson speaks freely of aesthetics in his account of the “discov- ery” of the double helix, often saying that the prettiness of the figures was an impor- tant standard of truth in the work of constructing a visual representation of DNA: James D. Watson, The Double Helix (1968; London: Penguin Books, 1997). this viewpoint, and the quote is intended to illustrate a general point rather than an individual statement. Subjective matters may be rele- vant when discussing art, but proper science is objective. William J. Mitchell describes this stance as being typical in scientific—and pho- tographic—procedures: Such exclusion of human bias is the point of many standard scientific proce- dures, such as random sampling, double-blind clinical trials, and setting the statistical significance levels before conducting experiments. . . . The photo- graphic procedure, like these scientific procedures, seems to provide a guaran- teed way of overcoming subjectivity and getting at the real truth.”28 Likewise, Lorraine Daston and Peter Galison write of scientists’ use of mechanically produced images in the late nineteenth century that, at issue was not only accuracy but morality as well: the all-too-human scien- tists must, as a matter of duty, restrain themselves from imposing their hopes, expectations, generalizations, aesthetics, even ordinary language on the image of nature. Where human self-discipline flagged, the machine would take over. Wary of human intervention between nature and representation, [scientists] turned to mechanically produced images to eliminate suspect mediation.”29 This—the alleged connection between mechanically produced images and objectivity—could explain why the researcher did not re- act positively to my attempts toward a more aesthetic discussion of his pictures, even though their aesthetic qualities are very much “in your face.” According to Galison and Daston, aesthetics and beauty are phenomena that are morally tainted in science, like subjectivity, and that the researchers should avoid—or at least ignore. Seen from this angle, it is not so strange that the young researcher tried to avoid discussions of something as “subjective” and “nonscientific” as colors and choice of motif. Still, the practice of making the pho- tographs legitimately entails both types of consideration: it is not ei- ther aesthetics or science; rather, it seems that science in this case could not exist without aesthetic considerations. COMSTAT: New Biofilm Representations After this discussion of the preparation of the CSLM-pictures and the experimental and biological “content” of these visualizations— that is, the color-marking and thereby the visualization of the bacteria 388 Configurations 28. William J. Mitchell, The Reconfigured Eye (Cambridge, Mass./London: MIT Press, 1992), p. 28. 29. Lorraine Daston and Peter Galison, “The Image of Objectivity,” Representations 40 (1992): 81–128, on p. 81. in the biofilm—I will go on to discuss the making and editing of the CSLM-photographs. However, before doing this I will consider an- other type of visual representation of the biofilms, namely COM- STAT-diagrams, because it makes analytical sense to consider COM- STAT at this point. COMSTAT is a piece of software that analyzes stacks of images produced by the CSLM, and produces diagrams to represent these analyses. As argued by Ferdinand de Saussure, signs acquire meaning not through their relation to reality, but rather through their relations to other signs. When I was trying to understand how to make and read CSLM-photographs, the researchers kept referring to COMSTAT and telling me that I should really look into those representations as well. This suggests that the two types of representation give meaning to each other through their difference. I asked the professor for his interpretation of the meaning of COMSTAT. He told me that COM- STAT was able to generate other types of representation of biofilms than the somewhat problematic CSLM–photographs: by measuring different parameters randomly across the biofilm, the software could generate quantitative rather than qualitative data. Thus, the more “subjective” (and thereby aesthetic) extra meanings of the pho- tographs would be countered by representations that were more tra- ditionally scientific.30 Sommerlund / Beauty and Bacteria 389 30. The sign is often classified into different types. Famous among these is the triad of Charles Sanders Peirce: icons, indices (often called “indexes”), and symbols. The iconic sign is the sign type I work with in this paper. It is characterized by resembling the con- ceptual object (“signified,” in Saussure’s terms); the usual examples are photographs, paintings, sculptures, and cinematic signs, but mathematical signs such as graphs and curves can also be put into this category. Peirce would say that both types of signs (pic- tures and graphs/curves) are icons, and thereby equally arbitrary. But even though After having spoken to the professor, I went on to talk to the re- searcher who worked on the COMSTAT software. I asked him why the COMSTAT program was so important. I will quote from this in- terview at some length: H: It [COMSTAT] is not as subjective as the CSLM-pictures. When you look at one of these biofilms, it shouldn’t be like ... you look at one biofilm and another biofilm, and then you go: nahhh, I think this one is a little bigger, and this one is a bit thicker and this one’s a bit thinner. If you do that, it becomes very subjective. Like, in some way, you know what you want . . . . JS: OK, so it’s the quality of the description that gets better? H: Yes, the quality gets better. The numbers in themselves aren’t that interesting at all. But the quality gets better because you dis- cover that maybe it wasn’t right what you thought you saw, be- cause you wanted to see something specific. JS: The CSLM-pictures are really difficult to look at, or so I think. Or maybe you get better . . . . H: It is difficult, and I think that one thing that is difficult about our way of doing research is that we’re really marked by our mod- els, or not models, but when looking at the biofilms we have strong hypotheses of why things are the way they are. And it is hard to look at them without having a hypothesis. It is difficult to look at them in a completely objective way . . . . So, the idea with this program is to quantify how these things look three-dimen- sionally. JS: But even with the program, there must be some prior under- standing . . . . There must be some boxes to put the results into? H: Sure. I decided what variables were to be calculated, like the thickness of the biofilm, and how big the surface is. Sure, that’s sort of a model that I have chosen. I chose to describe the biofilm this way. JS: How did you decide which variables and parameters to work with? H: I just took all I could think of, and then I ordered it according to rank. An immediate and intuitive thing: when something 390 Configurations Peirce—who worked as a chemist and mathematician as well as a semiotician—might regard CSLM-photographs and graphs as having the same status, the lack of color and the randomness of choice of motif made the graphs, curves, and numbers more palat- able to the microbiologists. grows on a surface, then how thick is it? The next thing that comes to mind is: how much does it vary, is it very flat or does it go up and down? That’s “roughness.” The next thing that comes to mind is: how big is the surface of this one, that’s important, be- cause how does fluid get in and out of this biofilm? And then there are a number of variables further down. The researcher refers directly to the “subjectivity” of the CSLM- pictures as constituting a problem. At the same time, he describes the outline of the COMSTAT program as neutral and objective, in spite of his “I chose to describe the biofilm this way.” His descrip- tions of the variables chosen as descriptive parameters of the COM- STAT-representation seem logical and obvious—however, it is also obvious that other variables could have been chosen, or they could have been ordered in a different way. Thus, it seems that the creator or designer of the representations is as present in the COMSTAT- graphs as he or she is in the CSLM-pictures. However, the quantifi- cation, abstraction, and generality of the COMSTAT-representations make them seem more scientific. Relevant to this point, Latour addresses the creation of diagrams and graphs representing the soil of the rain forest. He describes this process as simultaneous reduction and amplification: the researchers reduce materiality, locality, and multiplicity, and they amplify com- patibility, standardization, and so forth. He writes: “In losing the for- est, we win knowledge of it.”31 Likewise, in the COMSTAT-represen- tations, the researchers lose the biofilm, but win knowledge of it. In the eyes of the microbiologists of the Molecular Microbial Ecol- ogy Group, the graphs obviously belong to a different class of repre- sentation from the CSLM-photographs. I use the plural—“microbi- ologists”—deliberately, since this researcher’s views are repeated by the majority of the researchers in the group; I presented an early draft of this analysis to the group at a research seminar, and the re- sponse was unanimous: the COMSTAT representations are more scientific, because they are more objective, which again makes them more true. As Rheinberger writes, the sciences aim at true representa- tions of the world.32 In this case we have two, very different, repre- sentations of the same “world”; the question then becomes, which is truer? The two types of scientific representation are common to many branches of science, and may be categorized using Marie-José Sommerlund / Beauty and Bacteria 391 31. Latour, Pandora’s Hope (above, n. 7), pp. 38. 32. Rheinberger, Toward a History of Epistemic Things (above, n. 25), p. 102. Bertin’s terms graphisme (polysemic imagery, such as pictures and photographs open to multiple interpretations—in the present study, akin to CSLM-photographs) and graphique (monosemic data repre- sentation such as diagrams that have only one predetermined inter- pretation—here, COMSTAT-diagrams).33 Peter Galison presents a re- lated differentiation in Image and Logic, where he describes two types of visual scientific traditions. One, the image-tradition, makes visu- alizations that have as the ideal “images . . . that are presented . . . as mimetic—they purport to preserve the form of things as they occur in the world”; in the present study, these mimetic images can be compared to the CSLM-photographs, while the COMSTAT-diagrams can be seen as a part of the other tradition that is marked by an ideal of logic, which again visualizes using machines that are “counting (rather than picturing) . . . aggregate masses of data to make statisti- cal arguments for the existence of a particle or effect.”34 The latter class of scientific representation seems to be performed by a researcher who is an innocent medium rather than an active creator.35 Many trends within Science Studies (notable in this con- text is the feminism of, e.g., Donna Haraway) criticize this way of re- garding science and scientists: the sciences’ claim of seeing every- thing while remaining invisible is a “god trick,” a claim that the scientist’s position does not influence the knowledge produced. In the present study, the COMSTAT-representations of biofilm do not point as explicitly to the “author” as do the CSLM-photographs, but they are still dependent on authors and subjective choices, as the above interview clearly shows. They are not made by a “god,” and some “subject” other than the computer has chosen the parameters used for scanning the biofilm, even when computers do the scan- ning randomly. Thus aesthetics and form, in the sense of composi- tion, are included in the COMSTAT-representation to the same de- gree as they are in the CSLM-photographs.36 392 Configurations 33. Marie-José Bertin, La graphique et le traitement graphique de l’information (Pairs: Flam- marion, 1977). For further discussion, see Cambrosio and Keating, “Of Lymphocytes and Pixels” (above, n. 14). 34. Peter Galison, Image and Logic (Chicago: Chicago University Press, 1997), p. 19. 35. Daston and Galison, “Image of Objectivity” (above, n. 29), p. 81, note that “Let nature speak for itself” became the watchword of a new brand of scientific objectivity that emerged in the latter half of the nineteenth century. At issue was not only accu- racy but morality as well: the all-too-human scientices must, as a matter of duty, re- strain themselves from imposing their hopes, expectations, generalizations, aesthetics, even ordinary language on the image of nature. 36. It is also useful to compare to Cambrosio, Jacobi and Keating’s discussion, in “Arguing with Images” (above, n. 2), of “figuration” and “demonstration.” Ideally, Making and Editing CSLM-Photographs I now return to the analysis of CSLM photography, since the con- struction of sign systems and the creation of visualizations and sto- ries do not stop at the coloring of bacteria: the researchers also have to capture the colors on film and make them into photographs. This part of the representational practice requires as much work and con- sideration as the first one. The CSLM that makes the pictures of the biofilm is not one simple piece of equipment, but rather a cluster of machines taking up most of one room, and an extended network of apparatus. This was where I went with the researcher after he had killed the bacteria and in- jected the probes. The room containing the CSLM is very different from the benches where most laboratory work takes place: the latter benches are well lit, fully occupied by researchers and technicians, and bustling with small talk, music, and jokes; the room where the CSLM is placed is dark and quiet, for only one person at a time can work there (or two, if one is observing the other). The researcher and I started out by transporting the biofilm system on a trolley to the CSLM room. The researcher then disengaged glass chambers from the system, and placed them under the microscope. He looked, and asked if I wanted to have a look, too. While I was looking, he ex- plained that a laser makes it possible to see how the biofilm is orga- nized inside. However, he added, there is an important trade-off con- nected to this procedure: each time the laser cuts through the biofilm, it bleaches out some of the color. Thus, the more layers are cut, the better the picture; but the better the picture, the more the motif vanishes, making it impossible to produce any future pictures. When the researcher had found a good place on the biofilm, he took pictures, which could be seen momentarily on a computer screen to the right of us. He started editing the image right away; this, he said, is always necessary because the CSLM tends to blur the colors: “The CSLM does not register the colors clearly.” I looked into the microscope to check how the pictures were supposed to look— assuming that the image in the microscope would be the “original” Sommerlund / Beauty and Bacteria 393 “figuration” is the process that “concretizes notions” (p. 124) by giving them graphic form; this process can be likened to the work that lies behind the CSLM-images. “Demonstration” builds on figuration and partakes in “show-and-tell exercise [rather] than logical proof” (p. 126). In practice, these authors argue, these two types of visual- izations are superimposed and intertwined, as is also the case when regarding the CSLM/COMSTAT-representations. we were trying to re-create on the computer screen and subsequently on paper. However, it was very difficult for me to tell what was in- significant blur and what was significant variation. This was true for the color-tagging as well as the gfp markers: it was impossible for me even to establish how the original looked. I asked the researcher, who was sitting next to me editing the picture, how he could be sure that what he was doing on the screen was “correct,” when it was so difficult to see what the original looked like. Judging this, he an- swered, is a matter of practice. Having said that, he continued changing the blur into clarity: violets were changed into reds, grays into blacks, and turquoise into greens.37 The blurry quality of the line between reality and representation, and the interpretation necessary to decide whether turquoise “re- ally” means blue or green, are aspects that the researchers are very fa- miliar with. To illustrate this, I will quote another researcher from the Molecular Microbial Ecology Group: F: You have to touch it to understand it—I guess you do, too? It is a funny abstract way one makes it [the pictures]. You have some colored dots, and then you can make a lot out of that. It must be damned abstract for people coming from other areas. To me it is very simple . . . easy to understand . . . but I have to admit that it is difficult to explain. Especially to explain how we make conclu- sions . . . . It is hard to explain what we see from green and red. It has to do with legitimate interpretations. How much can you de- fend to put into those interpretations? There is a line . . . JS: How do you establish that line? F: It is something you learn from routine and experience . . . how much can you conclude from such a thing, and where is the line? Where do we stop? At which point can we no longer believe what we say? These are some of the hardest things. By referring to tacitness, routine, and experience, this researcher stresses the contingent nature of scientific work. In effect, this was also what the researcher I observed did, when he dismissed my questions as 394 Configurations 37. Samuel Edgerton and Michael Lynch notice something similar within astronomy: “What aesthetics means [in astronomy] is not a domain of beauty or expression which is detached from representational realism. Instead it is the very fabric of realism: the work of composing visible coherences, discriminating differences, consolidating enti- ties, and establishing evident relations. . . . This hands-on process of “interpretation” can be treated as an art situation within the performance of scientific practice” (Samuel Edgerton and Michael Lynch, “Aesthetics and Digital Image Processing,” in Picturing Power, ed. Gordon Fyfe and John Law [London/New York: Routledge, 1988], p. 212). based on lack of practice. This is very interesting, for the trait com- monly connected with hard science would be universality and not con- tingency. In regard to the CSLM-photographs, however, the researchers stress contingency as legitimizing the scientific status of the pho- tographs (and not, as might be expected, the opposite), thus again suggesting that science and aesthetics cannot be seen as opposites in connection with the CSLM-pictures.38 Likewise, the researchers stress interpretation as being a crucial task when working with the pictures. In- terpretation is traditionally connected to aesthetics and not to science, hence the concepts of “interpretive sciences” and “exact sciences.” This stresses that even references to universality and contingency, respec- tively, are contingent, and a part of the artisanal practices of science. The Extra Meanings of Photography Photography in itself bears many extra meanings that add to the meanings of the CSLM-photographs and influence their scientific status. A photograph (of, e.g., a bacterium) can be argued to be an icono- graphic sign: a signifier that refers to the signified by means of physical resemblance;39 interestingly, so is a graph.40 The signified in this case is a system of bacteria, which has been placed in a specific situation— a now—that has been immortalized on the filmstrip. In “Rhetoric of the Image,” Roland Barthes argues that photography holds a special status within sign systems, which originates in a direct link to the re- Sommerlund / Beauty and Bacteria 395 38. Karin Knorr Cetina, in Epistemic Cultures (Cambridge/London: Harvard University Press, 1999), notes that the senses play very different roles in the two branches of sci- ence studied, particle physics and molecular biology. In molecular biology, she argues, the sensory body is important as a secondary tool: materials are continuously visually inspected, and on this background the status of the experiment, photograph, etc. is de- termined. Likewise, the researchers studied by Knorr Cetina—and the ones studies by me—underlined the importance of “being there” in the laboratory. Even researchers with many junior researchers and laboratory technicians working for them insisted on doing important experiments “with their own hands”. 39. It has also been argued that pictures depict through denotative symbols. For fur- ther discussion, see Mitchell, Reconfigured Eye (above, n. 28), p. 24. 40. Rheinberger (Toward a History of Epistemic Things [above, n. 25], p. 103) notes that scientific representations follow the sign triad of Charles Sanders Peirce: symbols (analogies, hypothetical constructs), icons (models or simulations), and indices (ex- perimental realizations). It is interesting that both CSLM and COMSTAT representa- tions are icons (although their histories include different types of signs—e.g., indexes in the form of experiments), while at the same time they represent the two primary representational traditions pointed out by Peter Galison in Image and Logic (above, n. 34). This suggests that the type of sign is extremely plastic as regards content, and that differences in respect to attitudes toward them should be found elsewhere. ality it refers to; in comparison, other sign systems are completely cut off from reality—not referring to reality, but addressing that which we think of as reality in what seems to be naturalness.41 This characteristic of the photographic medium—that it holds a special status among sign systems because of a closer correspondence to reality—means that photography in some aspects resembles “hard science,” which also claims a special status (among epistemological systems) because of a closer correspondence to reality. This is com- mented on by Mitchell, who writes: “We feel the evidence it presents corresponds in some strong sense to reality, and (in accordance with correspondence theory of truth) that it is true because it does so.”42 He continues: “We have come to regard [photographs] not as pictures, but as formulae that metonymically evoke fragments of reality.”43 Thus, this characteristic of the photographic medium brings the photograph closer to the core of hard science: it is perceived as an objective, neu- tral investigation of something already existing. This is what Mitchell calls “the extraordinary tenacity of the camera’s claim to credibility.”44 Barthes´s claim that photography holds a privileged position when it comes to referring to reality can be disputed—and has been so by the later Barthes,45 among many others. For instance, W. J. T. Mitchell writes the following of the postmodern “pictoral turn”: Whatever the pictorial turn is, then, it should be clear that it is not a return to naïve mimesis, copy or correspondence theories of representation, or a re- newed metaphysics of pictorial “presence”: it is rather a postlinguistic, post- semiotic rediscovery of the picture as a complex interplay between visuality, apparatus, institutions, discourse, bodies, and figurality.46 On a less philosophical note, it has become obvious to all that the photographic medium is no more innocent than any other medium. Thus, it can be claimed that photography—both before and after the button is pressed—holds a number of possibilities of manipulation in regard to the CSLM-photographs. I regard this manipulation as being divided into three main categories: selection, manipulation, 396 Configurations 41. Roland Barthes, “Rhetoric of the Image,” in Image, Music, Text, ed. Stephen Heath (New York: Hill and Wang, 1977), pp. 32–51. 42. Mitchell, Reconfigured Eye, (above, n. 28), p. 24. 43. Ibid, p. 27. 44. Ibid. 45. Roland Barthes, Camera Lucida (London: Flamingo, 1981). 46. W. J. T. Mitchell, Picture Theory (Chicago/London: University of Chicago Press, 1994), p. 16. and editing.47 Through these examples of manipulation, photogra- phy is also removed from science and pushed toward aesthetics. First, the very selection of a motif can be seen as an instance of manipulation. Many photographs would not only look different, they would also mean something different if what has been left out were still in the frame, or vice versa. In relation to the work done in the Molecular Microbial Ecology Group, this is what most of the re- searchers stress as a problem when explaining to me the pros and cons of CSLM-pictures: it is only too obvious that an individual has selected the motif—it has not presented itself, and it has not been chosen randomly. Second, the motif itself can be manipulated. To say that the mo- tifs of the CSLM-pictures are heavily manipulated is no exaggera- tion: the natural setting that the pictures refer to has very little to do with glass chambers, tubes, and pumps in a laboratory. Likewise, the bacteria in the chambers are not the same as those they represent: the wild types are not blue or red, and they are not able to shine with a green light when their metabolic systems are activated. Third, much can be done during editing. In the rough end of the spectrum, shapes can be inserted into and removed from the picture; and in the more refined end, colors can be made brighter, fragments highlighted, and so forth. Editing the CSLM-photographs is a major task, the purpose of which is precisely to make the unclear become clear. In conclusion, the medium of photography contains forces that pull it toward the core of traditional science (the direct link to real- ity established through the physical resemblance to reality—that which Mitchell calls the correspondence-theory quality of pho- tographs), while others pull it toward aesthetics (selection, manipu- lation, editing). Thus, photography acquires a peculiar duplicity in its relation to reality: it can be perceived as the most “genuine” medium (when perceived in the commonsensical way of “Rhetoric of the Image”), but also as the most “fake” one (which should be ob- vious from the list of manipulations above). Thereby it also acquires a double status in relation to the scientific status of work that uses photography as representation and documentation. This point can be illustrated further through Galison and Daston’s discussion of “mechanical objectivity,” which analyzes how, histori- cally, the very mechanism of the mechanical representations was taken as a token of objectivity.48 This somewhat naïve belief that ob- Sommerlund / Beauty and Bacteria 397 47. William J. Mitchell’s discussion of “Intention and Artifice” in photographs (Recon- figured Eye [above, n. 28] pp. 23—59) suggests that this is characteristic of photographs in general. 48. Daston and Galison, “Image of Objectivity” (above, n. 29). jectivity can be ensured by mechanism has been tainted (we all know famous examples of “photos that lied”), and the photograph now holds a paradoxical double status in regard to subjectivity and objectivity: on the one hand, it is still seen as the most objective rep- resentative medium; and on the other hand, we all know that pho- tographs can be manipulated, chosen for the purpose, edited. . . . The researchers at the Molecular Microbial Ecology Group seem to perceive the forces that push the photograph toward aesthetics to be as powerful as the forces pulling toward science. In Galison’s book Image and Logic, he describes the “image-tradition” and the “logic- tradition” as equally strong within physics.49 If this goes for biology as well, this could explain the researchers’ need for representations from both traditions. In contrast, the layman seems to perceive photography as more reliable than graphs and diagrams. One way of understanding this difference between scientists and the rest of us is to consider the practical work that goes into the photographs, both in the sense of working with the motif (color-marking the bacteria) and in editing. It is obvious to anybody that these pictures are not the result of holding a snapshot camera over a lump of bacteria. This also means that technology is very visible in the photographs. They do not offer an easy way just to look at the bacteria represented; on the contrary, you are struck by questions such as How did they make this? How did they color the different strands? How is the 3-D effect achieved? It should be noted that these questions relate to the technology and work-practices, and not to the bacteria. Technology and practices end up being an implicit—but very important—motif, as is the case with works of art.50 This is underlined by the fact that some researchers ex- press great pleasure in technology: the visibility of technology means the visibility of practical skills—the mastering of that technology. You could say that the photographs function as starting points of different chains of signifiers: one connoting the explicit, scientific meaning of the pictures (Acinetobacter, P. putida, benzyl alcohol, gfp, metabolism, toluene-degrading), the other connoting the implicit, aesthetic meaning of the pictures (techniques [gene splicing, prob- ing], technology [CSLM], skill, artistry, creativity.) From this per- 398 Configurations 49. Galison, Image and Logic (above, n. 34). 50. I think of the CSLM-photography as a medium that refers to itself, just like certain traditions within visual arts. Traditions such as cubism point to the work of applying paint to the canvas, and the work of the eye in deciphering that paint and making it into a picture, as much as they point to the motif. The motif can almost be perceived as an “alibi,” and the result is a merger of representation and that which is represented. spective, the sign-chains are equal, and it is no longer possible to separate denotative and connotative meaning. Instead, both sign- chains become connotations, underlining how “denotation is not the first meaning, but pretends to be so; under this illusion, it is ul- timately no more than the last of the connotations.”51 Conclusion Aesthetics and science are intertwined in a very basic and practi- cal sense in the making of CSLM-photographs. At the same time, tra- ditional scientific representations (such as those produced by the COMSTAT program) are as constructed as the photographs. The es- sential difference to the researchers is that construction is not as vis- ible in the COMSTAT-representations, which makes the representa- tions seem less aesthetic and thus more scientific. The schism between aesthetics/construction and science/inves- tigation seems almost irrelevant when observing the practice of the researchers. Their practice utterly blurs the boundaries between the two and implodes the difference between representation and reality. However, there are still boundaries to consider; they do exist and do have practical consequences for the researchers. They exist as dis- courses within the scientific community and play decisive roles in many of the researchers’ practical representational choices. The sta- tus of the CSLM-photographs—as boundary objects between science and aesthetics, between which strong demarcations can also be found—is something of a paradox: they are both the flagship and eye-catching device of the group—its face to the world—and some- thing to be trivialized and joked about. Interestingly, through my narration of the story of the CSLM-im- ages and the COMSTAT-graphs, a series of concepts have been pre- sented—concepts that can themselves be regarded as elements in a narrative, and analyzed as such. The elements I refer to are placed in a dichotomous relation throughout the story.52 On the one side we find images, which were linked to aesthetics by the researchers in the beginning of the paper, and later to subjectivity by the same re- searchers. Subjectivity was linked to scientific practice, for no researcher would deny the presence of real and embodied researchers in the lab- oratory. Last, scientific practice was linked to narratives, first by the re- searcher who told the “real story” of the pictures of Pseudomonas putida, and second by me in narrating the story of the images. On a Sommerlund / Beauty and Bacteria 399 51. Roland Barthes, S/Z (London: Cape, 1974), p. 9. 52. In setting up dichotomous structures to analyze narration, I am inspired by Algir- das Julien Greminas, Strukturel Semantik (1966; Copenhagen: Gorgen, 1974). more general level, narrativity seems to be better able to capture practices, because it is stretched out in time as practices are, and it suits the type of real-time study that I have sought to conduct here. On the other side of the dichotomy, objectivity was linked to science —through the description of aesthetics and subjectivity as nonscien- tific. Thus, science was equated with “Science” and not with “science in practice,” which is obviously performed by real researchers in real time, and is hence deemed subjective. Science was then linked to truthful images because Scientific Truth is expressed in images (images performing the work of correspondence theory, presenting fragments of reality). The CSLM-images are the epitome of the efforts of the lab- oratory, the result of the purification work of inscription devices. It is extremely important to notice, though, that the point of this paper is not to point out some dichotomous relation between science and aesthetics. Rather, it is the opposite: the important thing to no- tice in the presented dichotomy is the presence of the image on both sides of the dichotomy. Here is an occasion of “slippage”: the “dou- ble nature” of the image underlines the point that a neat dichoto- mous split between science and aesthetics is never complete, because that which is cut out of science is also a prerequisite for science. In the laboratory in question, this double nature of the relation be- tween science and aesthetics means that the aesthetically beautiful CSLM-photographs function simultaneously as the laboratory’s crown jewel and as a back door through which all kinds of muddy, contingent, and “nonscientific” phenomena can slip in. Acknowledgments For insightful comments and readings, I would like to thank Signe Vikkelsø, the anonymous referees at Configurations, and editor Jim Bono. Also, I would like to thank the Molecular Microbial Ecology Group for treating me with courtesy and hospitality. The research for this paper was made possible by a grant from the Corrit-Foundation (Danish Technical University) and an internal grant from Copen- hagen Business School. 400 Configurations