Total primary inductance was actually determined by measuring the secondary inductance and then multiply it with the turn ratio squared. Imposed secondary voltage was then multiplied with the turn ratio, thereby yielding the primary voltage. Optional L(U(f)) dependency may be determined by regarding the fact that the mmf is proportional to the magnetizing current and thereby inversely reactance dependent at a certain fixed voltage. The voltage dependency at for instance "25Hz inductance" is therefore half that for 50Hz.
Comparing this result with that of the OPT, the L(U) dependency is more linear. But by viewing the L(I) dependency, this is at the expense of a much less DC current tolerance.
When considering an "Class A" OPT, it is important to note that the core must be able to withstand a certain small but considerable bias unbalance. One reason is tube aging differences, but a much more important reason is the dynamic matching even at pure Class A. Summarizing the requirements of the tubes and amplifier, the following should be fulfilled:
1) Static DC/bias-point matched tubes (or integrated bias balancing adjustment possibility)
2) Dynamic AC/cut-off matched tubes (important at high output levels)
3) Using tubes from the same production line (minimizes aging problems)
The above requirements is by the way always recommended but not necessary when using my special OPT TI-57613-C.