I have four of these units with the Ohms option. (Two are 8505A, two are 8506A, but everything related to Ohms is the same between the two models.)
I tested all four in parallel by hooking up radial wirewound resistors on all four units, 5 ohm and 0.25%, like this:
It seems very thermally stable, I could blow on the resistors and the readings did not deviate.
Here are the results:
2 hour warm-up. No averaging, all default settings. Peak mode turned on at start of test period to measure deviation.
4 wire Ohms mode.
Test 1: Peak to peak deviation in 10 minutes:
Unit 1 - 0.0005 ohms
Unit 2 - 0.0005 ohms
Unit 3 - 0.0001 ohms
Unit 4 - 0.0033 ohms
Test 2: Peak to peak deviation in 10 minutes:
Unit 1 - 0.0007 ohms
Unit 2 - 0.0005 ohms
Unit 3 - 0.0001 ohms
Unit 4 - 0.0046 ohms
Conclusion - Unit 3 is almost perfect, unit 4 is broken, while units 1 and 2 are probably "normal".
I have read somewhere (perhaps on the EEVblog) that the current generator for the 4 wire Ohms can become noisy on these meters. Maybe that is what is going on with my Unit 4. Perhaps the time has come to look at how that circuit works, and see if it can be made nice and quiet...