The peaks observed in \(G_1+ G_2\) vs. the dot levels \(\epsilon_i\) indicate the stability diagram for this setup. The tunneling rates are computed from the nonlinear differential conductances. From these measurements it seems that the tunneling rates \(\Gamma_i\) and \(\gamma_i\) are quite similar indeed. Measurements of \(G_2\) were insensitive to situations where the dot level in dot \(1\) was occupied or not. There are two striking features in the experiment:

• A drag current in dot \(2\) was detected, with \(I_2\approx 40 pA\) even thought the two tunneling rates \(\Gamma\), and \(\gamma\) were quite similar, in contradiction with the proposed theory.

• Even more striking the fact that for the situation where charge fluctuations in dot \(2\) were almost suppressed a drag current on dot 2 is observed. The drag current \(I_2\approx 0.5 pA\) decreases as \(-\epsilon_2\) grows.