We propose that the cause of the observed sinusoid in light curves of
EG Andromedae is ellipsoidal variation due to tidal distortion, demons
trate that the amplitude and wave form agree well with observations, a
nd analyse light curves to estimate fractional lobe filling by the red
giant component. We find 82 to 100 per cent lobe filling, depending o
n assumed values of inclination, mass ratio and rotation. If our basic
assumption is correct, the orbit must be far less eccentric than has
been advocated in some recent publications, and is likely to be circul
ar. The case of non-zero eccentricity fails by a wide margin to pass t
he Lucy-Sweeney test at the 5 per cent level of significance. We also
solve radial velocity data simultaneously with the light curves under
the circular orbit assumption. Either no rotation or synchronous rotat
ion of the red giant allows good solutions. However, sub-synchronous r
otation is a particularly interesting case because it may lead to mass
transfer without lobe filling and also explain EG And's erratic photo
metric and spectroscopic behaviour. If our proposed mechanism of varia
tion is correct, the red giant must be more than twice as large and mo
re than four times as luminous as previously estimated, and the binary
correspondingly more distant.