For two two bodies of mass M1 and M2 in circular orbits of radius a1, a2 about their common centre of mass, the Newtonian modification of Kepler's third law is a3/P2=G(M1+M2)/4π2. Where a=a1+a2. The problem is that I have been told that when using the units of years, solar masses and astronomical units, this reduces to a3/P2=M1+M2. I'm not sure how to show this is true, and find it quite strange that such a unit change could manage to perfectly cancel out G/4π2. The internet and textbooks don't seem to be very helpful about this so I was hoping somebody could point me in the right way, thanks!