A voltmeter, however good, always draws at least a tiny current to actually move its needle or display a reading — and the moment any current is drawn from a cell, the cell's own internal resistance causes a voltage drop, so what the voltmeter reads is slightly less than the cell's true EMF. A potentiometer sidesteps this problem entirely with a clever trick: it finds a condition where exactly zero current flows through the cell being measured, measuring EMF without disturbing it at all.
Setting up the potential gradient
A potentiometer is just a long, uniform resistance wire with a steady "driver" current pushed through it by a separate battery. Because the wire is uniform, the potential drops evenly along its length — a constant potential gradient k, in volts per metre.
Tap off the wire at a particular point and the potential there, measured from one end, is simply k multiplied by the distance from that end. This is the whole foundation of the method: distance along the wire becomes a direct, linear stand-in for voltage.
Finding the null point
Connect the unknown cell (EMF to be measured) through a galvanometer to a sliding jockey contact on the wire, with its negative terminal matched against the driver circuit's polarity at that end. Slide the jockey until the galvanometer shows exactly zero deflection — at that specific point, the potential drop along the wire between the end and the jockey exactly equals the unknown cell's EMF, which is precisely why no current flows: there's no potential difference left to drive any.
Why zero current at balance is the whole point
At any jockey position other than the exact balance point, some current does flow through the galvanometer branch, and that current flowing through the unknown cell's own internal resistance causes exactly the voltage-drop problem a voltmeter has. Only at the precise null point is the current zero, which is the only condition under which the wire's tapped potential genuinely equals the cell's true, undisturbed EMF rather than its terminal voltage under load. This is also why a potentiometer can compare two cells' EMFs directly and accurately (balance each against the same driver circuit and compare their balance lengths) in a way a voltmeter comparison would not do as precisely.
Where this shows up beyond the lab
The same null-deflection principle — adjust something until a detector reads exactly zero, then read off the adjustment itself — appears again in the Wheatstone bridge for measuring unknown resistance, and more generally any time a measurement is designed to avoid disturbing the thing being measured.
