We describe a new balloon-borne instrument (PoGOLite) capable of detecting 10% polarisation from 200mCrab point-like sources between 25 and 80keV in one 6 hour flight. Polarisation measurements in the soft gamma-ray band are expected to provide a powerful probe into high-energy emission mechanisms as well as the distribution of magnetic fields, radiation fields and interstellar matter. At present, only exploratory polarisation measurements have been carried out in the soft gamma-ray band. Reduction of the large background produced by cosmic-ray particles has been the biggest challenge. PoGOLite uses Compton scattering and photo-absorption in an array of 217 well-type phoswich detector cells made of plastic and BGO scintillators surrounded by a BGO anticoincidence shield and a thick polyethylene neutron shield. The narrow FOV (1.25msr) obtained with well-type phoswich detector technology and the use of thick background shields enhance the detected S/N ratio. Event selections based on recorded phototube waveforms and Compton kinematics reduce the background to that expected for a 40-100mCrab source between 25 and 50keV. A 6 hour observation on the Crab will differentiate between the Polar Cap/Slot Gap, Outer Gap, and Caustic models with greater than 5 sigma; and also cleanly identify the Compton reflection component in the Cygnus X-1 hard state. The first flight is planned for 2010 and long-duration flights from Sweden to Northern Canada are foreseen thereafter.

Linear polarization in X- and gamma-rays is an important diagnostic of many astrophysical sources, foremost giving information about their geometry, magnetic fields, and radiation mechanisms. However, very few X-ray polarization measurements have been made, and then only mono-energetic detections, whilst several objects are assumed to have energy dependent polarization signatures. In this paper we investigate whether detection of energy dependent polarization from cosmic sources is possible using the Compton technique, in particular with the proposed PoGOLite balloon-experiment, in the 25-100 keV range. We use Geant4 simulations of a PoGOLite model and input photon spectra based on Cygnus X-1 and accreting magnetic pulsars (100 mCrab). Effective observing times of 6 and 35 hours were simulated, corresponding to a standard and a long duration flight respectively. Both smooth and sharp energy variations of the polarization are investigated and compared to constant polarization signals using chi-square statistics. We can reject constant polarization, with energy, for the Cygnus X-1 spectrum (in the hard state), if the reflected component is assumed to be completely polarized, whereas the distinction cannot be made for weaker polarization. For the accreting pulsar, constant polarization can be rejected in the case of polarization in a narrow energy band with at least 50% polarization, and similarly for a negative step distribution from 30% to 0% polarization.