Brian O'Donnell

If normal mass is seen as a dent in space-time then -ve mass can be seen as a bump. From here there are 2 options. With Option 1, all test masses roll 'downhill'. For Option 2, +ve masses go 'downhill' and -ve masses go 'uphill'. In terms of F=ma, the inertial mass 'm' is the same sign and magnitude as the gravitational mass under Option 1. If you pull at a -ve mass it accelerates in the opposite direction. For Option 2, 'm' is always +ve, ie. |m|, and all masses accelerate in the direction of the force. Both options obey the laws of physics if you make consistent assumptions for each case. Option 1 follows Bondi's assumptions and has the advantage of inertial and gravitational mass being identical. Under Option 2, like masses attract and unlike masses repel with an inverse square of distance law, offering a kind of symmetry with electrical charges. Negative mass would end up in the spaces between the +ve mass galaxies, a possible explanation for at least some of the observed expansion of the universe.