This article draws on the theoretical insights of the archival turn and STS studies to reflect on ways to understand the natural archive and integrate it into future work in environmental history. I examine both herbarium specimens and ice cores to see how their creation and assembly into archives (the second kind of archival sampling error) creates a source that suggests certain narratives and can impede the creation of others. I argue that while historians’ use of the natural archive can broaden their available sources, the natural archive and the documentary archive are not fundamentally different: both are material traces of the past. This article argues that there is a grammar of archives whether the sources contained therein are old letters or old air bubbles: this grammar, marked by a syntax of omission and inflections built on uncertainty, is a result of the archiving process. Archives are not simply content; they are also a process done by humans. It’s this processing that makes the act of archiving dominate its contents. Perhaps even more importantly, rejecting the distinction between human history and natural history responds to recent calls to reunify these two histories and recast Homo sapiens as actors on a geologic scale. A well-theorised unified archive of documents and material objects will provide environmental historians with the undergirding for narratives of the past with greater depth and accuracy, and will be of value to responding to the challenges of the Anthropocene.