[volt-nuts] HP 3458A DC current accuracy

[Tony]

I'm new here so please be gentle with me if you think I'm asking stupid 
questions.

I'm trying to understand the accuracy specifications of the 3458A. One 
that I can't understand is why the DC current accuracy specs are much 
worse than the DC V and resistance accuracies would imply are possible.

Eg. The 100uA range has a 730 Ohm shunt resistor. Given that the figures 
for the 100mV DC and 1K Ohm ranges are 2ppm Reading + 3ppm Range and 2 + 
0.2 respectively, shouldn't it be possible to measure 100uA to an 
accuracy of ± 8.4 ppm from:

  (73mv + 2ppm * 73mV + 3ppm * 100mV  / (730 - 2ppm * 730 + 0.2ppm * 1K) 
= 100.00084uA

But the 24h 100uA DC current accuracy is quoted as 10ppm Reading + 6ppm 
Range, giving a maximum uncertainty of 16ppm when measuring 100uA. Why 
the big discrepancy?

The error due to self heating in the shunt resistor should be 
insignificant - looking at a relatively inexpensive ± 5ppm/K Vishay PT56 
for example, the temperature rise above ambient is approx 123K/W giving 
a maximum error of 5ppm * 123 * 100uA * 73mV = .0045ppm

Is it because the shunt resistance isn't measured as part of the meter's 
self calibration?

Actually I'm not that interested in the 3458A specifically, but more in 
the sources of errors in electrical measurements and the 3458A is 
specified very comprehensively.