In general, it's pretty easy to get the 8x or 19x correctly — these are the large dimensions. So really, you are only looking at being wrong on the 102, and off by two (100-104) is not such a big difference (1.52M-1.58M).

Once you realise that the error bars are small (and it was mostly intuitive for me, probably looking at counting up to a hundred, so a few percent off is not a big deal), you stop worrying about the uncertainty as much ;-)