relationship Po-Im-p /(IN), where Im-p is the peak amplitude of the mean IPSC.The weighted mean single-channel conductance (g) could be derived from the relationship gs-i/(Erev- Vh), where Erev is the reversal potential of glycine receptor- mediated chloride currents, and Vh the holding potential. In this study, the equilibrium potential of chloride ions was about-35 mV. Statistical analysis was performed using unpaired Students t test and data were represented as mcans+ SD 2(%7x(B…-4 Filtering properties of BCs under voltage-clamp recordings. revaluate the effect of filtering on the kinetics of IPSCs, we digitally- modeled butllfrog-retinal BCs with NEURON- software Hines and Carnevale. 1997) as following. Bullfrog BC has a soma of about 7 um in diameter, an axon of 50 um long and L um in diameter branching into two processes (30 um in length and 1 um in diameter), and a primary dendrite of 10 um long and 2 um in diameter branching into two thinner processes with 20 um long and l um in diameter/ (measured from BC images). Taking a typical input resistance for BCs of 10 G@2 when potassium channels were blocked during our experiments, we-can-calculate a membrane conductance per unit area of 0. 143 Sm 2wo .rind If-weAssumd a cytoplasm resistivity of 200 Qcm and a specific membrane capacitance ofIuFcm2, this would lead to a space constant()of about 935_um for 子 d.c.and about 197 um for 100 Hz a.c. value much longer than the typical length of4 BC dendrites and axon. For typical glycinergic IPSCs studied in present work with a 10-90% rise time t,of 1.5-3.5 ms a decay time constant(tp)of 10-40 ms and a peak amplitude (p)of 5-30 pA, T, and to will be slowed down less than 10% and 5% respectively, and I, will be decreased about 7% even when IPSCs were elicited from the end of BC dendrites or axon