Interesting. I think his interpretation is incorrect. The ANSI C57 standards he referenced talk about the "specified value" of impedance. When a transformer is ordered and manufactured, it will have a nominal impedance value specified and the standards have nominal impedance values. The ACTUAL impedance as measured by test can vary per the allowed range (generally +/- 7.5%). On older transformer nameplates, the impedance was left blank until the transformer was tested and then the actual value was stamped on the nameplate. I checked Conrad St. Pierre's book on short circuit calculations and he makes no mention of this issue in terms of adjusting nameplate impedances, and I've never heard it explained this way. Perhaps it is different with the very small dry-type transformers, but in my experience there is generally an actual impedance (not nominal) on the nameplate. (This is in the US.)