You need one or more known quantities.
Usually for analogue the calibration procedure is in the range of 0% and 100% scale optimised by turning the little screw until both are satisfactory (depending on the accuracy of the coils and the printer that made the papers it might be a trade off between either one perfect or both sort of near perfect).
A known quantity can be a very accurate voltage source, but for analogue mV scale will not be needed.
Tuning 0 is easy: Short the two terminals and hold it exactly level, while tuning.
Tuning full scale can be done with a power supply and a decent multimeter, or a power supply with a good enough display of its voltage output.
Then also see how off it is, and choosing what manner of averaging between the errors will work for you.
If it's a spiffy one that's been built for accuracy, you might want a mV scale and a very good eye and might be able to get both ends spot on.
EDIT:
Ah hang on, it says volts, but has no internal resistor, just a 1mA response coil.
So create a known current, by putting a resistor in series, so you can tune the voltage on your power supply to adjust the current, and again use another measurement interface to get to exactly 1mA. If you have some reasonably accurate resistors you could measure voltage across a 1k or 10k type and go for 1V or 10V. Measuring single volts is usually easier on affordable multimeters than measuring tenths of a mA.