We are used to weighing powder by the grain.
However brass is 8X the density of powder. 

8 grains of brass take up about the same volume as 1 grain of powder.

Now think about how large a variation in powder charge is the largest you will tolerate. 

For example you say you will tolerate a +/- .2 grain variation in powder charge.  

8 X .0002 = +/- 1.6 grains.  or 3.2 total variation

That is how I determine the limits of variation.
I weigh all the cases and see how many in the lot fit within a total spread of 3.2 grains. 

Another way is to weigh all the cases, record all the data and graph it.

Compute the standard deviation of the weights .
Then determine the limits of +3 sigma and -3 sigma

The cases in the tails of the curve would be considered the outliers to be used for working up loads, fouling shots and spares. Or the cases in the tails might make up a sub lot of heavy cases to remain grouped as a lot and a group of light cases to be grouped as a lot.

After you graph the data you can inspect the graph by Mark 1 eyeball looking for two different populations - the curve has 2 peaks (mixed brass from two different lots). 
You can go back and decide how big you want your bins to be based on visual inspection. Then resort the brass buy bins.

If you buy a large lot of loose brass not in sealed factory packages you can indentify if your brass is cull brass. Cull brass might come from a match shooter or OCD reloader. They might buy 1000 cases, weigh all of them, cherry pick the best brass out of the middle of the curve and then sell the brass that makes up the outliers.