Oil spills have taken on considerable political significance in recent years. The Exxon Valdez oil spill in 1989 was a case in point, since it resulted in the introduction of the first oil spill contingency plans as a precursor to extraction in the United States. In the summer of 2015 it was these very contingency plans that took center stage in the Arctic, as Royal Dutch Shell ventured north into the Chukchi Sea and critics questioned the oil major’s capacity to manage a spill in icy waters. How was the industry to govern an environmental disaster that had yet to occur? And how was it to speak for a material assemblage of oil, ice, and response technology that it had yet to encounter?

If the oil industry’s challenge was to investigate its capacity to respond to a potential Arctic oil spill event, then my challenge was to explore exactly how it did so. As Shell’s Polar Pioneer made its journey north, I set about talking to industry experts, government officials, and representatives of concerned environmental organizations. My aim was to get under the skin of the preparation process and to understand how an Arctic oil spill response capability had become the subject of such contestation. To borrow from Andrew Barry (2015), I found myself negotiating one particular “oil archive”—a body of knowledge created with the intent to push back the hydrocarbon frontier. And, as with any oil archive, this one contained both corporate science and counter-science, produced by the critics of the industry who sought to challenge its operations.

The corporate science to which I refer was grounded in laboratory experiments. These small-scale simulations worked to render the potential future of the Arctic oil spill present, and in doing so made it known and actable. In the words of Ben Anderson (2010), they provided a space for practices of performance, calculation, and imagination that would enable the industry to speak for its capacity to govern an oil spill in the icy waters of the Arctic. The Joint Industry Research Programmes responsible for the research deployed the full range of booms, skimmers, dispersants, and burning agents in controlled conditions in order to establish the effectiveness of an oil spill response in the presence of ice. Yet these small-scale simulations were in no way sufficient to speak for true Arctic conditions.

The oil industry needed to take its tests beyond the laboratory, but doing so would not be simple. Regulatory conditions in the Arctic are such that permits to spill oil are granted only in exceptional circumstances. However, in 2009 permits were acquired for a small number of field trials in Norway’s Barents Sea. For the industry, these tests served to clarify the findings of earlier experiments and would later underpin the contingency plans submitted to the United States. Certainly, there was a sense in the conversations I had with industry representatives that, in the case of Arctic field trials, the location facilitates and legitimates facticity. But there was a complex dynamic at play here, one that both enabled the verification of laboratory results and provided the grounds on which the oil industry’s critics would later challenge its claims.

Let me illustrate this argument through the example of oil weathering. One of the calculations carried out by the industry concerned the ignitability of oil that had been weathered in the presence of ice. At the laboratory scale this involved placing oil and prefabricated ice into a ten-meter metal flume housed in a room with a controlled temperature and wind speed (Sørstrøm et al. 2010). When it came to the field trial, the researchers travelled to Svea in Svalbard, where they cut a similarly shaped flume into the fjord ice. However, the similarities did not end there. They erected a tent and used heat generators and other apparatus to replicate the conditions present in the earlier small-scale experiments. It was in this way that the industry constructed its Arctic laboratory.

The outcome was akin to that which Geoffrey Bowker (1994) describes in his account of Schlumberger’s science during the early 1900s: just as aspects of the field were drawn into the laboratory for the initial experiments, aspects of the laboratory were taken into the field to ensure success and guarantee the commercial viability of the oil industry’s operations. In this case, however, the blurred boundaries between laboratory and field had not one, but two sets of consequences. First, they conferred upon the industry an ability to speak for the real Arctic environment, but second, they created an opportunity for the critics of the industry to question the validity of its claims. As a report commissioned by the Pew Environment Group noted, the field trials had more in common with a laboratory test than a real-life spill event, and this similarity provided the basis for counter-scientific claims to be advanced.

Bruno Latour (1983, 155) favored the metaphor of scientific facts as trains, writing that “you can extend the rails and connect them but you cannot drive a locomotive through a field.” In these terms we can think of the environmental movement challenging the capacity of the oil industry to extend those rails constructed in its laboratories into the uncertain Arctic future. Oil spill simulations have become central to the ability of petro-capitalists to push back the hydrocarbon frontier. Yet the very material character of these simulations has ensured that this process is far from straightforward.