I don't know what to say, guys; I can only report what I observe. However, I did screw up the decimal points in my original post. Where I was stating "0.5 grains" I was really intending 0.05 grains. I apologize.
I cleaned the bearings and knife edges with alcohol and a nylon brush (all I had at the moment), and checked the level of the bunch with a torpedo level, which, much to my surprise, showed the benchtop perfectly level. I couldn't believe it, so I rechecked it in several places-all perfectly level (bubble completely inside the lines.)
So I zeroed the scale again and noted the following results, all of which, you will notice are within the 1/10 grain margin of error that Redding claims.
The 0.5 grain weight reads right on 0. (or weighs 0.5 grains)
The 1.0 grain weight reads right on 0. (or weighs 1.0 grains)
The 5.0 grain weight reads right on 0. (or weighs 5.0 grains)
The 10 grain weight reads +0.05 (or weighs 10.05 grains, which it doesn't because it's a check weight, and I have nothing to verify the weight of the check weights, so I must assume that the weight weighs 10 grains, and the scale is off by +.05 grain)
The 15 grain weight reads +0.05 (or weighs 15.05 grains. if the 10 grn weight really is .05 grn too heavy, it would show here as well because I placed the 10 grn and 5 grn weights together on the scale.)
Here, I rechecked the zero of the scale by removing all weights and setting the counterpoises to zero. The scale returned to -0.05 (or the pan lost 0.05 grains of weight, or the scale simply doesn't return to zero.)
At this point, I re-zeroed the scale.
The 20 grain weight reads a hair over 0. (or weighs a hair over 20 grains)
The 30 grain weight reads right on 0. (or weighs 30 grains) (This tends to refute the idea that the 10 grn weight is too heavy and weighs 10.05, as discussed earlier, because, for 30 grains, I placed the 10 and 20 grain weights on the scale at the same time. If the 10 grn weight weighed 10.05 grns, then the 30 grn would have read 30.05 grns.)
The 40 grain weight reads a hair over 0. (or weighs a hair over 40 grains)
The 50 grain weight reads right on +0.1 (or weighs 50.1 grains)
All of this is within the 1/10 (0.1) grain margin of error specified by Redding, so cleaning the pivot points helped. I'm not sure sending this back will be of any help, unless it's to correct the not returning to zero issue.) Is it standard procedure to constantly re-zero these types of scales? Should I "zero" the scale with the check-weights on the pan, rathe than an empty pan? (In other words, I'f I know I'm loading 43.3 grains of powder, should I zero the scale with a 40 grn check weight in the pan and then start weighing? Ignore the empty pan zero?)