We consider the theory of charged point molecules ('probes') being pul
led by an electric field through a two-dimensional net of channels tha
t represents a piece of gel. Associated with the position in the net i
s a free energy of interaction between the probe and the net; this fre
e energy fluctuates randomly with the position of the probe in the net
. The free energy is intended to represent weak intractions between th
e probe and the gel, such as entropy associated with the restriction o
f the freedom of motion of the probe by the gel, or electrostatic inte
ractions between the probe and charges fixed to the gel. The free ener
gy can be thought of as a surface with the appearance of a rough, hill
y landscape spread over the net; the roughness is measured by the stan
dard deviation of the free-energy distribution. Two variations of the
model are examined: (1) the net is assumed to have all channels open,
or (2) only channels parallel to the electric field are open and all t
he cross-connecting channels are closed. Model (1) is more realistic b
ut presents a two-dimensional mathematical problem which can only be s
olved by slow iteration methods, while model (2) is less realistic but
presents a one-dimensional problem that can be reduced to simple quad
ratures and is easy to solve by numerical integration. In both models
the mobility of the probe decreases as the roughness parameter is incr
eased, but the effect is larger in the less realistic model (2) if the
same free-energy surface is used in both. The mobility in model (2) i
s reduced both by high points in the rough surface ('bumps') and by lo
w points ('traps'), while in model (1) only the traps are effective, s
ince the probes can flow around the bumps through the cross channels.
The mobility in model (2) can be made to agree with model (1) simply b
y cutting off the bumps of the surface. Thus the simple model (2) can
be used in place of the more realistic model (1) that is more difficul
t to compute.