This paper offers a detailed consideration of how the theoretical scope of the boundary object concept fits within an actor-network theory (ANT) sensibility when researching 'messy objects'. Messy objects are artefacts whose relational effects are inherently slippery and complex. As the aim of ANT is to show how non-human and human actors are co-constitutive in performing social activity, the uptake of boundary objects in ANT studies may appear to be an expedient analytical endeavour. However, scholars have raised concerns that the boundary object concept has lost some of its original analytical bearing, and that it is theoretically incompatible with ANT. This paper argues that a more careful reading of boundary objects' conceptual origins can provide useful insights for an ANT study. To illustrate this argument, findings are presented from an ethnographic study of engineers' knowledge practices in an emerging industry. Specifically, it shows that when a messy object--a signature on a contract--is foregrounded as a boundary object, particular knowledge practices are made visible. However, due to the complexity and messiness of the signature's performance, this paper contends that a pluralist theoretical approach to analysing messy objects may be more helpful to address issues of professional practice and knowledge.