The design of new protein-protein interfaces requires the ability to create reliably structured loops on protein surfaces. An attempt to change the conformation of an antigen-recognition loop on the surface of a therapeutic antibody will be described. While the loop took the desired structure, dimerization was an unintended consequence of the engineering. However, mutant information and a high-resolution crystal structure of the dimer provides detailed insight into why a destabilized antibody would dimerize and how this could be avoided. Modeling and molecular dynamics simulations in explicit water hint at an unfolding mechanism that may be generally relevant for antibody aggregation. Corresponding tryptophan fluorescence measurements provide additional support.