There are two things one can count on in Hans-Jörg Rheinberger’s books, including this latest one, Spalt und Fuge: Eine Phänomenologie des Experiments (Split and Joint: A Phenomenology of Experiment). First, they cultivate the reader’s sensibility for the phenomenology or “micro-epistemology” (Bachelard’s term) of experimentation. Second, they operate with philosophically provocative assumptions. Here, one of these provocations appears early on, with a critique of Latour’s concept of “technoscience” which Rheinberger takes to imply the identification of science and technology: “I will show that it is worth our while to distinguish rather than blur the technological and epistemological aspects of the sciences, and to study their interplay. This will render visible how the technical (Technisches) and the epistemic (Epistemisches) condition and promote each other—and how precisely through this they reaffirm the differences between them” (19).

It is tempting to dismiss Rheinberger’s point by arguing that the term “technoscience” is nowadays used to do just that—to distinguish science and technoscience by tending to their different ways of configuring the technical and the epistemic (e.g., Bensaude-Vincent et al. 2011). If what one means by “knowledge” is something like a true theory or accurate representation of phenomena, then it will be evident also to philosophers of technoscience that there are technical achievements which do not produce new knowledge, especially when they can be modeled or explained by means of extant theory. And if this is what one means by knowledge, theory development and hypothesis-testing rely on technology only instrumentally or merely as a tool, subservient to representational aims: There would thus be an interplay between the distinct spheres of the technical and epistemic. But according to philosophers of technoscience what one means by “knowledge” might also refer to how one gets to know the world through building and making, disclosing what things can do as they are artfully put together to produce novel effects. On this Baconian conception, there are technical ways of knowing and thus a conflation of the technical and the epistemic. Whether or not this conflation exists marks a difference between two different epistemic projects, and it seems odd that a philosophy of experimentation aligns itself with only the first of these.

To be sure, Rheinberger’s reasoning is more nuanced and subtle. Readers of his earlier work on epistemic things and technical objects will recall that, for him, the sphere of “the epistemic” includes everything that is of intellectual interest, whereas “the technical” includes everything that is settled or under control (Rheinberger 1997). Now there is a temptation to say that science seeks to bring everything under control and render things technical. According to Max Weber, the aim of science is to “master everything by means of calculation” and thereby to advance the disenchantment of the world (Weber 1946). Weber’s remark is echoed by Martin Heidegger‘s account of modern science as representing all things in terms of cause-effect relations to render them amenable for technical control (Heidegger 1976). On this account, modern science recruits all of nature for technology and thus promotes a technological approach to and understanding of things. According to Latour, this exposes the conceit that there are separate spheres of science and technology, nature and culture in the first place (Latour 1993).

In contrast, Rheinberger’s account is much closer to that of Charles Sanders Peirce: The work of the mind is provoked by intellectual interest and seeks the settlement of questions which leads to the formation of habit and controlled regularity. However, this settlement is not final but opens up new matters of intellectual interest (e.g., Peirce 1992, pp. 304–305). Our instruments and experimental procedures serve not only to stabilize phenomena, they also register difference and irregularity. It is thus the very effort to transform questionable and interesting epistemic things into reliably predictable technical objects which continually produces new epistemic things. The regime of technical control is also a sensorium for the discovery of scientific phenomena. Rheinberger’s micro-epistemology of the significant traces that inhabit the crevices and intermediary sites of experimental practice thus proves to be cutting-edge and nostalgic at once. His careful attention to neglected aspects of experimentation is coupled with an idealized one-sided view of what scientific or technoscientific research is all about. This is evident early in the book in a discussion of bioinformatics. After noting that “All experimentation finally amounts to the production of traces” (22), he remarks with a note of regret that bioinformatics transforms these traces into data: “When treated as data, traces become tied down and disambiguated. They are thus deprived of the precarious, the tentative, the preliminary, the degree of indeterminacy that attaches to them” (32). And yet, “behind the space of data and traces” that is produced by sensors and analytic chip-technologies, “there persists the world of epistemic things, the knowledge of which is the point of all these procedures” (34).

This is where the strengths and weaknesses of Rheinberger’s program come together. His Peircean account of an ever more technically determined world behind which opens up another world of unsettled problems and questions yields a veritable manual of categories and fine-grained distinctions that help describe experiments as machines for the production of epistemic things. Tied up with this analytical tool-kit is an unquestioned assumption, namely that this is what experimental research is all about and that this “is the point” of all experimental procedures—to arouse intellectual interest, to feed curiosity, to seek out epistemic things. Rheinberger’s rich conception of experimental practice thus preempts the question what kind of practice it is or what it is a practice of. Where Hasok Chang argues for the centrality of intentionality and directedness of practice (Chang 2014), Rheinberger does not question whether and how experimental practice is oriented to a particular conception of knowledge—whether to achievements of control or to the production of traces that grow into theoretical questions. This striking omission is no blind spot, however, but a profound commitment that is firmly entrenched in the circular logic of Rheinberger’s argument:

But most of all I want to maintain that a modern researcher who in an empirically detailed manner pursues and advances a science cannot be a—theory-driven—engineer but rather resembles a bricoleur. If a segment of today‘s synthetic biology appeals to an engineering spirit, this leads it mightily astray (23).

If pursuing and advancing a science is categorically different from engineering, such researchers evidently can’t be engineers. Anyone working in an empirically detailed manner will produce epistemic things and expose limits of theoretical knowledge. In contrast, engineers, including many synthetic biologists, seek to generate complexity by bracketing the unknown, by relying on modularized bits of theory, by trying to uncouple explanation and understanding from reliable prediction (Nordmann 2014, 2020). By being interested in the unknown, the scientist can only be a scientist and not an engineer or technoscientist. Fair enough. But by being interested in what one can do with what is already known, the engineers or technoscientists also investigate the world in empirical detail, also expand knowledge, and might even be producing epistemic things though these are not what they are primarily looking for. Rheinberger’s narrow focus makes sense since he wants to show how experiments produce epistemic things, but it comes at the unnecessary and unreasonable price of underestimating and devaluing technoscientific research or “an engineering way of being in science” (Galison 2017).

It is for his consideration of experiments as machines for the production of epistemic things that Rheinberger’s book deserves to be read. The book’s title already refers to the interstitial, invoking the cracks and joints and the recalcitrance of things which refuse to be seamlessly integrated in a hegemonic, totalizing, homogeneous account, be it a theoretical picture of the world, or be it a comprehensive working order of things. His account follows a tradition that includes Gaston Bachelard and Deleuze and Guattari, as well as Walter Benjamin and Theodor Adorno, Nancy Cartwright and John Dupré. To their rather principled ways of advocating for the things, Rheinberger adds a “phenomenological archeology” and builds a tool-box and terminology for tracing distinctive strategies of dealing with and at the same time rendering salient epistemically relevant things. Here Rheinberger is a most effective translator who brings in theorists who are not usually recognized by philosophers and historians of science. In this volume, these include anthropologist Claude Lévi-Strauss and art historian George Kubler. With Kubler, Rheinberger opens a discussion of temporal forms of experimentation, according to which experimental achievements build on each other or, rather, grow from each other as do artworks, forming sequences where things lead us from one to the other (146). The unit of analysis is not a Kuhnian paradigm but rather the work of art, the technological device, or experimental achievement as a temporary resting point not in an evolutionary development but in a series of temporally contiguous things that leave an imprint on each other—a series that might result, for example, from grafting as a bioexperimental practice that transforms experimental systems through hybridization (103). Rheinberger distinguishes this temporal dynamics or developmental process from the temporal form of discontinuous trajectories where a certain way of doing experiments comes to an end, replaced by other instrumental and procedural modalities. It is distinct also from the temporal development of experiments themselves, in the laboratory or the field, unfolding in the manner of gradual elaboration and articulation or in the manner of dramatic surprise. And it is distinct from the work that goes into detemporalizing the experiment and of rendering it timelessly reproducible. By being interested in the often neglected middle range of temporal forms, in hybridization by grafting and in the unfolding of sequences, Rheinberger trains the historical sensibility of his readers and opens new fields of questioning.

The phenomenology of experimentation proposes how things appear and how they draw attention, nothing but traces of traces that serve as epistemic things, far removed from becoming technical objects. Through the lens of experiment as Rheinberger conceives it, the world appears as neither a totality of facts, nor a totality of things, but a collection of fragments. The book ends with the whirl of elements that emanates from the dynamic shape of a nautilus shell and inspires the work of artists like Albert Flocon or Mikhail Vrubel. Bachelard and Rheinberger tend to the material basis of a shell as an origin of the imaginative motion that unfolds in the work of an artist. Similarly, they tend to the experimental system as a site which gives rise to a temporal dynamic by which—in the final words of the book—“knowledge can be wrested from the world only in waves which expose and consume, which split (spalten) and join (fügen)” (254). And thus, tending to highly specialized, technologically sophisticated experimental systems that operate in a strongly application-oriented research setting, Rheinberger arrives at a romanticized image of the scientist. It is quite literally “romanticized,” perhaps heuristically, as evidenced by the book’s last main chapter on “fragments.”

By the end of the book, “all experimentation amounts to the production of traces” has become transmuted into something like “all experimentation splits the world into fragments and fragments are epistemic things par excellence” (cf. 240). With the ethics and epistemology of fragments Rheinberger refers us back to the romanticism of Goethe, August Wilhelm and Friedrich Schlegel, or Novalis. The fragment in this tradition is a melancholic reminder of a lost whole, a hint or cue for its reconstitution but also an aesthetic form in its own right, a cypher for reflection about history and agency. Like Rheinberger’s epistemic things, it beckons for integration into a larger picture of the world, and by way of its obstinacy it simultaneously embodies the impossibility of any such integration. If experimental cultures serve to split and fragment the world, explore fragmented sections of reality, and produce fragmentary traces, this speaks to the endless struggles, the unfinished business of science:

And especially the experiment, the central procedure of science, is fragmentary by its very nature. Fragmenting here becomes the requirement of an entire mode of knowledge. […] The fragmentation of reality for the purpose of its analysis is the basic procedure of empirical research (238).

Rheinberger’s conservative provocation thus manifests its heuristic power: What used to be discussed under the heading of experimental control, the isolation of causal processes, the creation of phenomena that signify stable features of reality, he reinterprets as the aesthetic production of fragments. This reinterpretation is a reconciliation with none other than Goethe who is known for his severe critique of modern experimental science (241). By agreeing that experiments torture and fragment reality, Rheinberger concedes Goethe’s polemic and gains from this a fruitful view of the relation between models and experiment: Models take up fragmentary experimental findings and seek to reconstitute them within an imagined whole. This, to be sure, is a precarious undertaking which requires constant qualification as experimental observations correct and amend each other. This process does not issue in the one theory or model that compellingly ties the fragments into a seamless whole. Instead, Rheinberger appears to adopt a morphological view of experimentation which produces a series of fragmentary exemplars that might form a sequence where one is an intermediary of the other, with the sequence pointing significantly in some direction.

Goethe as a critic of science thus helps Rheinberger argue not only for the close relation of science and art but also that of scientific philosophy and cultural history: Their business is equally precarious as both need to reassemble a fragmented world—with the one difference that archeologists build knowledge from the ruins of the past, whereas the process of fragmentation is actively pursued by means of the experimental method of empirical science (251). First, the world is split into innumerable facts, and then all these bewildering facts need to be made sense of and tentatively reassembled into some picture of the world.

At a time when philosophers of science and of technoscience are interested in resilience and robustness, in scientific understanding beyond mere explainability, in predictive control of intellectually intractable systems, Rheinberger posits a sublime view of a subtle and precarious enterprise, an intellectual adventure fraught with epistemic risk, with temporary stability at best in light of the manifold fine-grained adjustments and attunements required to manage that risk. And thus, the very prejudice that leads him to valorize a certain view of science opens the door to his thick description, a profound and inspiring reading of experimental practice.