2025-2026
(di)atomic garden is an artwork in the form of a real-time, performative system. It explores moments of resonance between how radioactivity acts as a mutative force, combined with it's witting/unwitting relevance for the worlds ocean in both the historic context of nuclear contamination, as well as the ongoing reframing of the oceans through agri-industrial lensing. Here the atomic garden is tasked as a bridge between terrestrial and the oceanic as their borders are uneasily dissolved.
Radioactivity has a powerful, altering effect on genomic material. Here the work explores how two disparate datasets of images (historical agricultural crops and antarctic marine plankton) as colliding gene pools mutated through the virtual simulation of an atomic garden. Starting in with digitised signals from a geiger counter monitoring a small quantity of uranium ore: the emission of charged particles by decaying uranium ore are sampled, and then used to drive a virtual re-enactment of the atomic garden. Each specimen planted in this virtual space is a careful translation of images into pseudo-genetic code and algorithmic encapsulation. This new, virtual form lends them vulnerable to mutation when irradiated by charged beams of radiation, leading them to be mutate and interact with each-other. The planted specimens can be explored through the browser interface that acts as a safe window into this continuously growing, mutating and decaying virtual garden.
The project draws on the little-known history of atomic gardens – experimental fields developed after the Second World War to research “peaceful” applications of nuclear energy. By exposing crops to radioactive materials, researchers induced mutations that sometimes resulted in higher yields, new colours, or novel forms. Remarkably, some of these varieties still appear in contemporary seed catalogues. At the same time nuclear weapons testing, reactor accidents, and industrial activities introduced radioactive residues into the world’s oceans. As a result, many marine organisms—from plankton to fish—now live with chronic, low-level exposure. While the effects of high radiation doses are well studied, the long-term ecological consequences of low doses remain uncertain, particularly when combined with ocean warming and acidification. These unresolved questions form the conceptual backbone of (di)atomic garden.
For as long as nuclear technology has existed, it has carried with it the issue of contamination: the controlled (and uncontrolled) release of radioactive material into the environment. We often tend to focus on singular, dramatic events such as Chernobyl, Fukushima or the detonation of nuclear weapons. Yet these are only the most visible instances within a much broader history. Radioactive contamination is routine and geographically widespread, often folded into ordinary industrial, military, and scientific practice. In the case of the ocean, this history includes not only testing and accidents, but deliberate disposal, quiet leakage, long-term infrastructural neglect, and the persistent assumption that the sea can absorb what land cannot.
Even an incomplete map of ocean-related cases in which radioactive material has entered marine environments makes this visible. What appears is not a sequence of isolated incidents, but a cultural and political pattern: the ocean repeatedly treated as a repository for unwanted matter, buffered by depth, distance, and the hope of dilution.
It remains difficult to draw firm conclusions about the full impact of these different forms of contamination. In many cases, the material dumped decades ago remains out of sight and difficult to monitor, while corrosion, dispersal, and ecological uptake unfold over timescales that exceed most human systems of observation. Some may already have unfolded unnoticed; others may still lie ahead. The countless barrels of radioactive waste discarded at sea are among the clearest expressions of this suspended uncertainty.
What can be said with confidence, however, is that the quantities involved are far from negligible. The volumes of radioactive material leaked, dumped, and released through detonation are substantial by any objective standard. It is therefore reasonable to assume that the conditions for unintended consequences have existed, and continue to exist, across many marine environments. The ocean, in all its scale and multiplicity, is not only a site of life and circulation, but also, in part, a lingering gamma garden.
It is from within this context that the work’s virtual garden takes shape. The garden is structured around two elements: a mutative force and a field of planted specimens arranged around it. The mutative force is radiation, but not as inspiration alone. Whilst anything can be simulated, there is a very interesting potential for the integration of non-typical (i.e. physical) inputs into virtual systems. Why? One repeating point-of-view has been informing much of our recent work, that the divide between what we consider digital vs. physical is an entirely artificial divide. Computation starts and remains the product of rocks and metals organised in specific (and admittedly very clever) ways.
For this reason, the mutative force driving the virtual garden is sourced from actual radioactive decay. Harmless, low-potency uranium ore was coupled with a custom setup involving a Geiger counter, electronics, and a live data pipeline capable of streaming the real-time detection of charged particle emissions into the system as virtual emissions. What unfolds onscreen is therefore not simply a simulation of irradiation, but a virtual ecology continuously perturbed by real radioactive events. Each mutation is the result of actual emissions.
(early prototype with geiger counter, digitising circuitry, and uranium ore vial. As emissions are recorded from vial of radioactive ore on the right, they are emitted virtually (left) into the (di)atomic garden)
If radiation provides the garden’s mutative pressure, the specimens provide its vulnerable bodies. These specimens originate in two source datasets: one drawn from historical agricultural surveys, the other from Antarctic phytoplankton imagery. Yet the specimens themselves are not simply images placed into a virtual environment. Rather, they are images transformed into another operational form. Building on ideas explored in the ongoing series 'self-contained', each source image is converted into a string of pseudo-genetic code using the bases G, A, T, & C. This encoding is intentionally naive, but sufficient for the work’s purposes: the image as mutable simulacra of life in turn makes it capable of corruption, inheritance, and mutative variation.
Alongside this encoded sequence, each specimen contains its own decoding apparatus. Using the original image as a reference, a purpose-built neural network (loosely inspired by the ribosome) is trained to reconstruct the specimen from its pseudo-genetic code back into pixels. Each specimen therefore consists of both a “DNA” sequence and a trained decoder contained within its own structure. The images encountered in the garden, including their subsequent mutations, are generated through this relationship. Although the work begins from image datasets, those original images are effectively replaced by specimens: image-forms reconstituted through encoded sequences and neural translation.
This technical structure is not incidental. By shifting images into a form susceptible to mutation, they become open to a different class of interactions. DNA, whether literal or metaphorically staged, invites thinking in terms of corruption, drift, replication, death, and propagation. Within the garden, specimens do not remain fixed. They mutate, decay, and are replanted over time, producing an ongoing and unstable negotiation between variation and selection. The work asks what kinds of mutations become legible as useful, interesting, viable, or beautiful, and how such judgments shift when two highly different image lineages are brought into contact under mutative pressure.
The choice of these two datasets, historical pomological illustrations and Antarctic phytoplankton, is deliberate. At first glance, the pairing may appear arbitrary. But their juxtaposition offers a way of thinking through the increasingly entangled relation between terrestrial agricultural logics and oceanic life. Historically, agriculture has been framed as a terrestrial enterprise, while the sea has more often been approached as a zone of extraction through hunting, transit, or resource capture. That distinction is weakening. As agri-industrial practices increasingly extend into marine environments, the ocean is more and more reframed through the lens of cultivation, management, productivity, and yield. This shift often remains abstract or difficult to picture. Bringing these two image worlds into the same mutative garden gives that merging a speculative visual form.
The pomological material is drawn from the USDA’s Pomological Watercolor Collection, produced between 1886 and 1942. Commissioned by the Division of Pomology, the survey documented an extraordinary variety of fruits and nuts, including countless cultivars bred, named, and catalogued under an explicitly productive regime. The collection is scientifically systematic, but also visually striking in its attention to variation and abundance. It speaks to a long history of ordering the more-than-human world through utility, classification, and improvement, while at the same time revealing the lushness and specificity of cultivated difference.
Set against this are the phytoplankton images, courtesy of Martina Mascioni, Allison Cusick, and the FjordPhyto Participatory Science initiative. These images were produced from samples collected in Antarctica by participatory scientists aboard tour vessels, as part of an ongoing effort since 2016. The dataset not only brings microscopic marine life into visibility, but also foregrounds the value of citizen science and distributed observation in environments that are otherwise difficult to monitor closely. If the pomological survey reflects a historical apparatus of state-led agricultural categorisation, the phytoplankton dataset emerges from a collaborative practice of environmental witnessing in a changing ocean.
Placing these datasets into the same garden allows their boundaries to erode. Fruits, seeds, cells, shells, membranes, and blooms begin to share a common mutational space. The garden stages an uneasy convergence between terrestrial cultivation and marine ecology, not as a literal prediction, but as a way of thinking through the consequences of human activity that increasingly dissolves their separation. As the system continues to grow, mutate, and decay, it explores the possibility that the productive logics once imposed primarily on land are now seeping into the ocean, while contamination itself operates as another agent of entanglement.
Taken together, these elements form a context in which the garden can be understood not simply as a digital image system, but as a performative ecology shaped by contamination, translation, and exposure. It is animated by real radioactive decay, populated by specimens that are no longer quite images, and structured by the collision of two archival worlds whose relation is becoming newly urgent. The work does not attempt to resolve the uncertainties surrounding radiation in marine environments, nor to stabilise what mutation means within such a setting. Instead, it inhabits those uncertainties—technical, ecological, and historical—and gives them form as a living field of transformation.
2026 ORBIT_E virtual extension of MBAL (CH) (link)
