Hi
Ok, the lowest cost way to do this is to buy a nice fresh battery. Measure it's voltage before use. Better yet, buy a pack of 10 of them. Measure all of them and pick one. Check all your meters against that battery and what it should read. The result is going to be pretty close to the accuracy of the gear you are using.
A year from now buy another pack of AA cells (or whatever you use a lot of) and repeat the process. Check against the same "what it should read" number. You can now maintain calibration against the same "standard".
Net cost ... zero if you use up the batteries in all of your battery powered gizmos.
Next lower cost:
One resistor, one cheap NPN transistor, one cheap 6.2V zener diode. The transistor gets wired with collector and base shorted. Wire them in series, bring out two leads. One is from the "ground" leg of the zener. The other is from the junction between the resistor and the collector / base combo on the transistor. The emitter of the transistor just goes to the cathode of the zener and does not come out to the outside. Any fairly good metal film 1% resistor will do. You want a miliamp or three through the contraption with a source picked in the 12 to 24V range powering everything.
You now have a sub 50 cent temperature compensated reference.
Grab your bench supply and set it to some voltage (let's say 12V, yes the resistor would have to be the right value). B+ goes to the top of the resistor. Ground goes to a separate lead to the anode of the zener. Set it with a good DVM borrowed from a friend. Read the voltage on the zener after everything has stabilized. Check the room temperature and record it. Also record the supply voltage and zener voltage. Go have a cup of coffee. Repeat the process in a half hour. Bump the supply up and down 2% and record that effect (just so you know).
You now have a cheap reference that is much better than your DVM's. To use it, you will have a little back and forth between the power supply set and the zener reading (it's a ratio and meters are pretty good at that). The trick is to read both the zener and the supply on the same range of the meter. That drops out most calibration issues on the meter. In use, you should be within a couple of mv (call it 10 mv max) of the recorded value. 0.1% of 6.8V is 6.8mV.
For added precision, run the beast at home for 30 days before you take it over to your friends house. For even more fun, pot the parts inside a piece of copper tube. Ground the tube to the supply ground lead. Scrap tube from a plumbing job will do just fine. A scrap is a sub 25 cent item if you have to buy it at the hardware store.
Yes you can go a bit crazy building a number of these things and temperature compensating them. You can also have four of them and check one against the other to be sure you are using one that has not broken. You can enhance the thing by using a 78L12 to power the resistor. That gets rid of most of the power supply dependance and could easily double the cost to (yikes!!) $1.
Like any project, this can quickly spiral off into enormous un-needed complexity (and lots more fun).
Bob