That isn't the "limitations of both meters influencing the other". Rather, that's the extra resistance of the measurement meter slightly reducing the current in the circuit to the DUT.
Let's take a specific example. I took my voltage reference which has a 5.00377 volt output and connected my 100 ohm 0.005% resistor (Y1453100R000V9L) in series with it. On my Agilent 34401A meter I measured 47.4017 mA. Under identical circumstances my AN8008 measured 49.9mA! That's because the shunt resistance on the 34401A is 5.022 ohms as compared to 0.0301 ohms on the AN8008. If you put both meters in series I would expect both meters to read about 47.4 mA due to the 34401A's shunt resistance. Which method is more meaningful and/or accurate? I want to know how accurate the current measurement is compared to the current without the meter inserted (50.0377 mA).
In the above situation the AN8008 was more accurate than my relatively expensive 34401A. However, if you correct the 34401A's reading, its accuracy is -0.016% as compared to the AN8008's accuracy of
-0.2452%. I'm trying to find the meter which has the best accuracy without requiring me to use my "slide rule."
Sorry, I feel like I'm clearly missing something really obvious.
Why does the actual current matter? Why go to the effort of using a precise voltage and a precise resistance in the hope of achieve a particular current when the total resistance added by the meter will be affected by the fuse (which is non-linear) and connectors/wires in addition to the shunt?
Or to put it another way, what is wrong with what I'm doing?
As I say, I don't care about the actual value of current - I just want to see how it compares to my known-good meters at certain approximate points, and to achieve that goal, I can't see any flaws with putting two or more meters in series. No slide rule required.
A current meter is specified in terms of the current passed through it, not the reading that results when some arbitrary voltage source is connected to it via some arbitrary resistance, no matter how precise those might be.
Obviously any current meter will have a shunt (and fuse, etc), so the current will be affected when you insert it into a circuit. The value of this resistance is just a fact of life, and is hardly a "figure of merit" given that all current meters must have a shunt to work. Just about any measurement will change things to a small extent; it's up to the operator to consider the effect of the particular shunt given the context: often it simply doesn't matter; other times, the burden voltage might indeed be a problem. But all that is quite separate from determining the basic accuracy of a current meter.