Noparking, you should find in your E-P1 manual the way to choose the metering pattern: something like "average", "center-weighted", or "matrix" (where "matrix" can be called "pattern" or something to that effect).
Then you should choose "center-weighted" or "average". The default choice is probably "matrix".

Auto-exposure (full auto, that is normally called "Program" mode or whatever P in PASM means) or Aperture priority do not have any influence in how the light reading is performed so there should not be any difference in exposure between the two automatic modes (or any other exposure mode).

When reading the light reflected by the bloody grey card a small different in the angle between card and light meter (or camera) matters. In your first example you got a 1/3 EV of difference (f/8 and f/9) while in the second example you had the same reading between incident and reflected light reading. I suppose the first time you did not have the exact same angle, or maybe you were projecting some shadow by mistake.

Be wary when doing this kind of experiments of the shadow you, the photographer, project on the grey card or on the light meter. It's easy to make this mistake. You'll see that if you repeat your measures with great care, you'll probably obtain very coherent results.

I don't use grey cards any more. The "caveats" in their use are many: angle between light source, camera, and light meter, mistakes due to shadows, mistakes due to "mirror effects" (reflectivity of the card at certain angles) and general clumsiness and difficulty in use. Incident metering works MUCH better and is MUCH faster