POLARIMETRIC ACCURACY REQUIRED FOR THE DETERMINATION OF BINARY INCLINATIONS

An analysis is made of the required accuracy of measurement of the linear polarization Stokes parameters (Q, U) variations from close binary stars in order to obtain a useful estimate of the orbital inclination i, based on the analytic canonical model involving single Thomson scattering in a corotating circumstellar envelope. The procedure comprises evaluation of the confidence interval of i for which the model yields an acceptable χ2 fit to simulated (Q, U) data when optimized over the other free parameters involved.In particular, results are presented for the maximum polarimetric standard deviation, σnec permissible on observations for the determination of i to ± 5° in this model at a significance of 10 per cent. The results depend on the true binary inclination, and (weakly) on the geometry.In particular the required accuracy of polarimetric observations is found to be significantly greater for low true inclinations.As an example of the application of our computations the results are applied to data for the binary U Sagittae. It is found that the size of the error on the polarimetric Q, U data, after phase binning of the observations, is some five times larger than the maximum error allowing satisfactory inclination determination for a true inclination of > 60°. This factor increases rapidly as lower values of i are considered.Possible complications are discussed which arise when the data contains noise due to intrinsic variations in the binary system (i.e. non-corotation, long period changes in the scattering geometry). © 1981 Royal Astronomical Society.