Stresses in a clamped circular plate
First I'll address the semantic issue: you are absolutely correct that the word traverse means across, no more, no less. However, if you told someone that traverse means across, handed them a washer, and asked them what the traverse dimension is, my bet is they would tell you it's the outer diameter; across the entire shape.
As for checking my results, as I stated in my last post, I compared an FEA model's boundary results to the formula's values and they correlated VERY closely. The key here is that my diameter/annulus ration was about 10, so the answers would have been very different if Roark was using the annulus dimension as the traverse dimension.
I'll state it clearer than I did last time: without a doubt, the meaning of "least traverse dimension" in Roark's for this application is the least out-to-out dimension. Again, as I said before, go set up a quick FEA model to prove it yourself. Seriously, it takes maybe 20 minutes to check it. You'll see what I mean. Roll up your sleeves and get dirty - it's by far the most effective learning tool...