The following was done using an RCBS Uniflow powder measure with an activated linkage in a Lee Turret Press. I used Accura #5 powder.
I had the rotation activation in so to get each powder measure the turret would do a full rotation. I used a Hornady GS-1500 scale with a resolution of 0.1 grains, in a room with some AC air blowing around.
For the 230 grain bullet at this load the range between starting and do not exceed is:
Jacketed : 7.8 grains to 8.7 grains.
Lead Cast : 7.7 grains to 8.5 grains.
Intersect : 7.8 grains to 8.5 grains.
First I tried the rig without any baffle:
Within 99% 8.54< X <9.33
3 sigma 99.73% 8.47< X <9.39
Within 99.99% 8.34< X <9.53
99.999999% 8.06< X <9.81
Then I cleared out the powder dispenser and put a powder baffle in:
No changes to the powder settings.
Within 99% 7.86< X <8.38
3 sigma 99.73% 7.82< X <8.42
Within 99.99% 7.73< X <8.51
99.999999% 7.54< X <8.70
The baffle has a definite improvement in performance.
(Note any numbers beyond a tenth of a grain are purely speculative given the limits of my measuring instrument)
Though the accuracy still leaves something to be desired.
The gents on GBC have some suggestions that I'll agree with. It could be the dispenser. A Dillion was recommended. It could be the the scale. It could be the airflow from the vents (using a three sided box with a top was recommended and is a simple test).
One thing is important, a scale with better resolution would be quite helpful. As it is now I'm just spitballing.
Though the difference between baffle and no baffle is at least large enough for me to see.
Next up is doing the enclosure test to see what that gives me.
Edit: Realized I goofed on my percent band calculations. Corrected