That was my point ;-)
I've just been checking up, and the weak link is the panel meter. I can get a 0.1% current reference built for well under £10, but a 0.1% panel meter starts at £17. That's one-off prices, but these things are not mass-market and there will be a significant mark-up. Unless you bought an especially high-accuracy meter, you almost certainly have something accurate to around 1% of full-scale. Assuming it reads to 19.99 ohms (because it's got to be at least 3, and the rest is free) that's 0.2 ohms. If there's a 1.999 ohm range, then it will be accurate to 0.02 ohms.
That's assuming all components except the meter are more accurate, so the meter accuracy predominates. Murphy's Law states "In an instrument or
device characterized by a number of plus-or-minus errors, the total error will be the sum of all the errors adding in the same direction." That would at least triple the inaccuracy, but it isn't generally that bad.