When you use a winding stick you lay it across the board. If the board is curved or otherwise out of true in that cross/transverse axis then you'll get a bad reading. That's pretty much true of all measurement approaches, though - local variations can always muck with larger-scale measurements one way or another.
Going back to my previous post and its replies, you would be well served to listen to Warren. He has more experience than anybody else here that I know of with hand-tool-only workflows. There is usually more than one way to do any given operation, and IMO like most humans he sometimes overstates the advantages of his approaches w.r.t. others, but you can pretty much bank on the fact that whatever he suggests WILL work very efficiently and well. In your situation it's going to be a lot more useful to have such "tried and true" approaches at your disposal than to try to understand all of the variations.