This article takes scientific ‘raw data’ as its ethnographic object in order to investigate the co-implication of nature and culture in scientific knowledge practices. The article traces out some of the activities that are involved in producing numerical climate data from the Brazilian Amazon. Although science and technology studies (STS) makes a strong case for associating relationality with certainty, the article argues that a particular form of data, ‘raw data’, complicates this association. It further argues that scientific data is not simply composed out of relations, but is a relation itself. The article ends with a brief reflection on the possible repercussions of shifting from thinking of science as producing multiple natures and cultures to thinking of it as producing the potential for relations.