Fig. 6. Slower recovery than the prediction of the one-site model when lidocaine is present in the drug mixture. (A ) In drug mixtures containing 100 μm drug, it is evident that the actual (experimental) recovery speed is well predicted by the one-site model when there are two anticonvulsants but no lidocaine in the mixture (thin white bars ; 107 ± 5%, 90 ± 8%, and 85 ± 7% of prediction, and n = 7, 6, and 5 for carbamazepine plus lamotrigine, lamotrigine plus phenytoin, and phenytoin plus carbamazepine, respectively). For the derivation of predicted recovery speed, see Results. In contrast, the actual recovery speed is always slower than the one-site model prediction when lidocaine is present (thin gray bars ; 72 ± 5%, 74 ± 5%, and 72 ± 5% of prediction, and n = 5, 5, and 4, for carbamazepine plus lidocaine, lamotrigine plus lidocaine, and phenytoin plus lidocaine, respectively). All data from the experiments where only the anticonvulsants are present are together to give an average “% predicted speed” of 94.8 ± 4.2% (n = 18, thick white bar ), whereas pooled data from the experiments with lidocaine in the drug mixture give an evidently smaller “% predicted speed” (72.6 ± 2.6%, n = 14, thick gray bar ). ***P < 0.005 by Student t test (compared between the pooled data of the two-anticonvulsants experiments and the anticonvulsant + lidocaine experiments). (B ) Larger deviation from the one-site model prediction with higher concentration of lidocaine in the drug mixtures. The percentage of the predicted speeds are 93 ± 1%, 69 ± 8%, and 65 ± 1% for 200 μm carbamazepine plus 200 μm lamotrigine, 200 μm lamotrigine plus 200 μm lidocaine, and 200 μm caramazepine plus 200 μm lidocaine, respectively. n = 3 for each column. *P < 0.05 and ***P < 0.005 by Student t test (when compared with the data of the two-anticonvulsants, that is, carbamazepine + lamotrigine, experiments).