I have and love my Harrell's. With ball powder and consistency of operation, I've found it will easily keep within one-tenth grain. With H4831, you'll have a three-tenths variation. With IMR4895 and H4895, the variation is down to less than two-tenths grain.
The question then becomes: Does that variation make a difference in the real world? I ask that because on at least two occasions I have run tests using thrown charges versus individually-weighed charges. In both tests, the thrown charges produced a slightly smaller Extreme Spread over an Oehler 30P chronograph than the weighed charges.
To be conclusive, of course, you would have to run the test many more times with different powders, etc. None-the-less, I no longer worry about small deviations. I will, of course, re-run the test when my new long-range rifle is finished.
Cheers,
Richard