I did some testing yesterday - I found variations up to .2 grams for identical slabs - see http://coins.aplawrence.com/2014/05/tare-weight-of-pcgs-slabsaws-100.html That's a lot of variance.
For larger coins is it ?? A 1 oz. American Eagle weighs in at about 33.93 grams, even off 0.2 would be a pretty accurate double-check, no ?
Don't know what the tolerance is for an AGE, but for a double eagle which is close to the weight of the AGE, the tolerance was +/- .03 grams. A .2 gram +/- range for the slab swamps the range for the double eagle. A fake DE could be five tolerances off and due to the tolerance range of the slab still be undetectable.
You would have to crack the coins out of those slabs to get meaningful data. You are showing a variation of .2 grams for one generation slab and .09 for another but the nickels in the holders have a tolerance of +/- .194 grams. So even if the slab shells weighed exactly the same it would still be possible to have as much as .388 grams difference between two slabs with nickels in them.
SLAB WEIGHTS: I decided to get some of my money's worth out of that digital scale I bought years ago and did some random measuring. For Saints, I found differing weights by about +/- 0.2 grams. Same slab type and same era of grading, so pretty much apples-to-apples. For silver coins like Brittania's or Isle of Man's, I found the variance to be +/- 0.6 or 0.4 grams. I don't have that many coins out that have duplicates where I can compare measurments. Maybe I'll dig out some more and report back. Obviously, I have no idea if the coin or the slab is changing in weight (I strongly believe the latter) but as the scale appears accurate to 0.01 grams, I guess the question is if a total weight (coin + slab) that changes by 1/2 or 1/4 gram is within normal tolerance limits.