The battery is needed because it supplies the voltage that is placed across the resistor being measured (after being divided between an internal decade resistor and the resistor being tested). The meter actually measures this voltage, and shows it on the non-linear resistance scale.
http://www.angelfire.com/electronic/...ubes/VTVM.html
The operation to measure resistance has two set-up steps
1) with the test leads shorted, adjust the meter zero adjustment
2) with the test leads open, adjust the ohms calibration to read infinite resistance - this is calibrating the reading when the meter sees full battery voltage.
3) connect the external resistor (Rext) to be measured.
The voltage measured is the battery voltage (full scale) times (Rext/(Rext + Rint)) where Rint is the internal decade resistor. The non-linear scale is marked to indicate the value of Rext. For example, if Rext = 333k and Rint = 1M, the voltage measured will be battery times 1/4, so the 1/4 of full scale point is marked as .333 (Megohm); the half-full-scale point is marked 1 (1 Megohm); etc.