Maybe there's some good way to do this- if not, here's some ideas:
1) Hunt up the construction drawings/ data from 1942.
2) Crawl all over the thing measuring dimensions and thicknesses using ultrasonic gauge, and calculate weights from that.
3) If you have some way to put a measured amount of fluid in the tank, I think you could deduce the weight from the displacement and float depth. The measured amount of fluid could be from another tank. My thinking is, you fill up the tank enough to completely cover the floor, then measure the additional depth required to float the roof. The difference in depth less the difference in volume would give you displacement of the roof, which should equate to the weight. You'd have to account for the minor contribution of the legs which were submerged partially at the beginning. Whether this would be any easier or more accurate than #2, I can't say.
Two other things- check that for sure this is required. For example, if you are calculating volumes by the change in depth, you don't need that weight- it's only when you drain the tank down past the floater that you need it. Secondly, if you haven't, run this question past the calibration people- I can't imagine this is the first time it's come up.
