Context : There is growing evidence that a treatment of binarity amongst OB stars is essential for a full theory of stellar evolution .
However the binary properties of massive stars - frequency , mass ratio & orbital separation - are still poorly constrained .

Aims : In order to address this shortcoming we have undertaken a multiepoch spectroscopic study of the stellar population of the young massive cluster Westerlund 1 .
In this paper we present an investigation into the nature of the dusty Wolf-Rayet star and candidate binary W239 .

Methods : To accomplish this we have utilised our spectroscopic data in conjunction with multi-year optical and near-IR photometric observations in order to search for binary signatures .
Comparison of these data to synthetic non-LTE model atmosphere spectra were used to derive the fundamental properties of the WC9 primary .

Results : We found W239 to have an orbital period of only \sim 5.05 days , making it one of the most compact WC binaries yet identified .
Analysis of the long term near-IR lightcurve reveals a significant flare between 2004-6 .
We interpret this as evidence for a third massive stellar component in the system in a long period ( > 6 yr ) , eccentric orbit , with dust production occuring at periastron leading to the flare .
The presence of a near-IR excess characteristic of hot ( \sim 1300 K ) dust at every epoch is consistent with the expectation that the subset of persistent dust forming WC stars are short ( < 1 yr ) period binaries , although confirmation will require further observations .
Non-LTE model atmosphere analysis of the spectrum reveals the physical properties of the WC9 component to be fully consistent with other Galactic examples .

Conclusions : The simultaneous presence of both short period Wolf-Rayet binaries and cool hypergiants within Wd 1 provides compelling evidence for a bifurcation in the post-Main Sequence evolution of massive stars due to binarity .
Short period O+OB binaries will evolve directly to the Wolf-Rayet phase , either due to an episode of binary mediated mass loss - likely via case A mass transfer or a contact configuration - or via chemically homogenous evolution .
Conversely , long period binaries and single stars will instead undergo a red loop across the HR diagram via a cool hypergiant phase .
Future analysis of the full spectroscopic dataset for Wd 1 will constrain the proportion of massive stars experiencing each pathway ; hence quantifying the importance of binarity in massive stellar evolution up to and beyond supernova and the resultant production of relativistic remnants and X-ray binaries .