Use of epoxy for levelling

[Neale]

Here is the corresponding error plot for my rail (as I have only managed to fix down one rail so far. Once visitors have left...)



I have converted to mm for ease of comparison. We used different resins, although I'm inclined to dismiss that one as a significant factor - they are both at the runny end of the epoxy viscosity scale. However, I was working at an ambient temperature of around 10C at the time of pouring and although I had kept the resin indoors for a few days beforehand, it was a thin film sitting on steel and will have been down to garage ambient almost immediately. That won't have done wonders for viscosity and maybe it didn't flow as well. I think my rail section error was also rather greater to start with, and as my errors are roughly aligned with the initial error (the bends in the rail) I wonder if these are linked. Better flow where the epoxy is deeper?
There is also the measurement technique to consider. I don't have a surface plate big enough to be useful - mine's a sheet of thick plate glass about 300mm square - so I used a level at a series of points, combining level scale readings with feeler gauges to get a "difference in height" value. I dropped all these into a spreadsheet to get the graph shown here. It wouldn't be difficult to calculate shim values to bring the rail into line. Is it worth it? Assuming, obviously, that the bed is finished to the rails, then a hump in the middle wouldn't be a problem as the workpiece would tend to droop over it, being clamped at the edges. However, it is more likely to bridge a hollow in the bed. And I'm quite certain that my sophisticated woodworking techniques would dramatically fail with an error of 0.18mm in depth over a 1.8m length...