To put it politely, my phone just died in the middle of a long post about impedance and L or C measurements, and I lost all that.
Basically, the short version (some may prefer, anyway), is that Z or L measurements for iron-core components, particularly power devices, vary widely with measurement method for may reasons. After much frustrating comparison, I engaged a transformer mfr. in discussion about this. Basically, measurement at conditions far from actual operating conditions will produce different measurements than expected. Recommended an incremental inductance analyzer like General Radio used to make. It typically is used with a power source (tube amp in the original form). I had suspected excitation/magnetization current cannot be provided by a small-signal L or LC 'meter' or Z analyzer.
Using a variety of such instruments, I too observed that iron core inductance increased at a fixed frequency between low and highest-available signal level (usually no more than 10 Vrms). I also saw variation with frequency, vaguely recalling higher Z at lower f. These observations were with ungapped transformers (power and output, EI and toroid core).
Knowing I only achieve a relative measurement, I use winding inductance measurements with a grain of salt. I kind of trust leakage inductance measurements for each winding because it's done by shorting a transformer winding and it's theoretically independent of the core. Anyway, since each shorted winding in a multi-winding component produces a different LL, it's relative info only. If you have a transformer with a persistent short (not high-voltage dependent), a full or partial winding short will cause inductance measurements closer to leakage inductance than 'actual' (permeability-dependent).
I compare pri/sec L at a single frequency due to proportionality to Z ratio, then square root of that as a first dart throw to determine turns ratio and identify windings (unknown components with no or insufficient data). I sometimes then apply voltage with a Variac/Powerstat (VAT) for another dart throw at turns ratio. They are usually far enough different to be disappointing. Comparing those to small-signal Z/L instruments show other significant differences. It at least helps me identify windings and how balanced dual primaries are regarding L, R and unloaded transformed voltage (turns) ratio.
I have also once or twice set up a 500 VA variable (amplitude & freq) AC power source to compare primary winding current as a wishful indicator of winding impedance (resistance+reactance...R+XL)...also frustratingly 'off' from other collected superstitious experimental measurements...probably includes current due to impedance and magnetizing current...all of which led to pinging a transformer engineer who hasn't changed his e-mail address or ignored me yet.
I haven't yet bothered to add series resistances, or attempt DC-biased L measurements...and won't, since my 'dry' AC-only data has been minimally useful.
I am satisfied with what little it does to identify/confirm a few things.
I guess not only frustrating but humbling and educational...showing me to not blindly take measurements as gospel.
Measuring old electrolytics with 3 different types of instruments is another 'educational' pursuit. Trust nothing until you can confirm.