Bright sub-mm and FIR sources have historically been counted amongst the most star-forming objects in the early universe. However, unambiguously identifying a physical counterpart at other wavelengths (in particular, at the optical, where dust extinction is usually severe) has proven very difficult. In this talk, I will discuss new results from a recent project, wherein we make use of the availability of higher resolution, shorter wavelength data to identify the locations of potential counterparts to the FIR detection. Using Bayesian inference methods via the XID+ software, we determine the best distribution of FIR flux, given the positions of the shorter-wavelength sources. In this work I address specifically the question of how frequently these FIR-bright objects are composed of multiple FIR-flux emitting components.