I tried with many different capacitors and resistance with a consistant behavior (that is, the cap in the triplet with the large resistance was observed to a lightly less voltage than the other cap). I also tried with relatively small caps+resistors with the same behavior (but with less voltage difference).
It seems that the error in the readings is consistently larger with the larger resistances, which could explain the observed difference though, which was something that I hadn't considered before !
Thanks for you contribution, it is always nice to see that someone could be interested in your problem.