Dan:

You now understand the difference between reality and science. Engineers deal in reality. Science is for people who can't handle reality, -- mathematicians, physicists and the like... everything works out in "physics land" -- speaking relatively of course.

OK when you measure "Zero" connect the two input leads on the phidgets. Then there "should" be "Zero" difference -- in the Engineering sense where everything is expressed so:

The input was 0V +- .0001 V

The power supply was built to a tolerance of +-.01 volts using precision components of 1% tolerance...

The voltmeter is accurate +- 2.5% of full scale...

...and statements like that

Go to Sparkfun (.com) -- look up some sensors and read the data sheets... For the pressure and temperature sensors sometimes your head swims... See the BMP085 at Sparkfun for example... That is what I am working on today. The BMA 180 tomorrow...

Get the idea?

**** When you add or subtract the absolute error (estimate or tolerance) adds absolutely...

**** When you multiply or divide the relative error adds absolutely.

Absolute error --> The output was 10V where I can measure to an accuracy of -.1v to + .2V volt of the value measured -- is a absolute measurement and error range.

Relative error --> I measure the output to be 9.41V where the meter was on a 10V scale with an accuracy of 2% of FULL SCALE.

So 2.25V (+-2%) / 100ma (+- 5%) gives you what resistance?

where R = E/I (Or V/I) if you like... (Volts over amps)

I leave it to the interested reader....

Error analysis is a weird and wonderful topic.... Anytime you are in the analog world, think component and measurement tolerance and the above methods...

So why do the "meters" disagree? You tell me -- but first pull out the manual and check the accuracy -- which I bet will be given as a percentage (relative) of full scale.

The pull out the arduino specs for your processor and find the A/D conversion accuracy specification. Are they relative or absolute?

When a commercial product is designed you calculate all these possible errors (somehow) then you try to get enough measurements to verify that your calculations are correct.

Just for fun look up the SPICE design program for circuitry.

Go from there...

And I'm just joking about mathematicians -- I work in a branch where the only way you can solve a problem is develop a well thought out opinion of how to get the answer... see "NP" and "NP hard" if you care... Then you develop guestimates of how to do the calculations -- then you pray that your clients don't catch on... It has been called witchcraft not mathematics by some engineers I know. I work in both areas and can reconcile this easily. (see Doublethink)

BTW: You will get your propeller beany in the mail... just attach it to your hat!