You also often have to sort it by lot.
Here's a link whose data includes two lots of Winchester 243 cases, both with 54.8 grains average case water overflow capacity and one lot with an average weight of 158.58 grains and the other with an average weight of 166.44 grains, despite that identical internal capacity. I, too, have noticed Winchester's weight going up since they moved the plant from Alton, Illinois. I don't know if it's because they changed tooling, or if it's because they are outsourcing some of it and the tooling style used by the new supplier is different.
The thing to bear in mind is that the thickness and diameter of the head, the rim and the extractor groove depth and relief angles all have tolerances that affect the case weight, but not the interior volume. So tooling changes can affect results.
Wm. C. Davis showed a long time ago that for most rifle cartridges, when the heads were identical, you had to have about 16 grains of case weight difference to need one grain difference in powder charge. I've found some powders for which it's more like 14 grains, so I use that as a worst case number. Both numbers assume cases made of 70:30 cartridge brass,
which they all are not, complicating things a little with their alloy density differences.
Worst case, if you have a good load that can tolerate at least half a grain of charge variation without running off its sweet spot, you are looking at ±3.5 grains of weight variation either side of the case weight it was developed in before you risk sending it off the end of the sweet spot range. If there are other factors, like changing temperatures or sloppy charge dispensing that you need to have that insensitivity range as a buffer for, then you might want to hold your case weights a little tighter. But sorting them to tighter than a grain and a half of case weight or about 0.2 grains of water capacity is not worth spending much time on, as a 0.1 grains of powder is all the equivalent difference it will make.
Incidentally, in .30-06, 0.2 grains of water capacity is about 0.3% of total case volume. So that's the level of precision you want from a volume measuring method for that cartridge. Water is the easiest way to get it, as long as you keep in mind that
water density varies with temperature and allow for it when you need to. When water is just about hot enough to boil at sea level, it expands almost 4.2%, which is the water density the volume of a fluid ounce is based on.
I expect the reason the OP wants the resolution he does is to be able to see differences that small between two cases quickly, even if the absolute number isn't true.