there's been some good info in this thread....but by just using 5, 6, 7 or 8 meshes alone won't guarantee you that will produce the "pure" as there will always be contaminants the same size as the heads that remain and are not completely separated later
the main things to keep in mind is to not allow the contaminants in, rather than try to remove them later..so always do at least two runs, one gentle run for the pure and then another more aggressive run for the remainding resin heads on the material using a wider range of meshes that will never be as clean (no matter what you do...as has been noticed in this thread)
and the next thing to do on this first "pure" run is to isolate your desired resin heads within a narrow range, rather than sifting all of the various head sizes and then trying to purify them...do this by firstly working out what your desired range of resin heads are within eg a 10 or 20 micron range and then start dry sifting by using two meshes that are close to the max. and min. of this range (not by using 220, 160 or other micron meshes that will allow much more non-active rubbish in with the good stuff)..then clean the heads on the minimum mesh with a card forcing the contaminants and smaller heads through (while the selected range remain on top of the mesh)
The above method will produce a much purer hash than techniques previously posted (anywhere lol) ...I also suspect that this is how Sam Skunkman produces his legendary dry sift using his "secret method" (as well as by using the best clone selected from 40 years of growing, aged plant material and by sifting it between a very narrow select size range of meshes)
and of course, resift all of the remaining material, including what was left above and below the first run to reap the remaining heads and then clean it using techniques mentioned in this thread - multiple meshes, agitating, carding and or vaccuuming etc etc)
happy dry sifting!