orbital mechanics question

I have a question about how drag affects are calculated.
Analytically, you start with rp=a(1-e), take derivatives and solve to
get edot=adot(1-e)/a because rp is constant (radius of periapse).
However, ra (radius of apoapse) is also constant. Doing the same
calculations you end up with
edot=adot(1+e)/a The semiparameter is constant also, and will give a
third equation. The first equation is the right one because of years
of success in practical applications; but the other two solutions are
just as valid analytically. If you want an explanation read my "f and
g functions" theory at www.redcolony.com

Bill Clark, P.E.