This paper responds to a growing literature arguing that change in higher education institutions might be better understood and managed if such institutions are understood as complex systems with emergent properties. Based on complexity theory, the paper articulates a set of characteristics that might be expected if institutions are in fact acting as complex systems. The utility of these characteristics for identifying complexity in the field is tested in the context of a large partnered research project involving the mathematics, science and education colleges from six Australian universities and their local K-12 schools. The study finds evidence of subsystem variations on the initial partnership priorities, including substantial boundary crossing, leading to emergence and co-evolution, indicating that a macroscopic view of emergent variation rather than 'micromanaging' may be an essential factor in scaling and sustaining collaborative partnerships.