It took all of half an hour today to rig up a butt tester. It is not pretty but I am confident in its rigidity and hence accuracy.
The support bar deflects .03mm with the full weight of a tube on the end but actual error with the tube weight mostly supported by hand is insignificant. The pip on the support bar is turned to a point to prevent 'bridge' error.
I have uploaded a couple of pics, see below, sorry they are a bit big!

I have 3 reynolds 31.7 tubes and 2 x 35mm tubes all are supposed to be .8 .5 .8

The 35mm tubes are most definitely .9 .6 .9 as is one of the 31.7 tubes. The other 31.7 tubes average about .85 .55 .85 but seem to be quite variable. Butt lengths and transition lengths look fine.

These findings explain the greater than expected tube weights.