Robert Baer wrote:

Some 12Au7 tubes have "unbalanced" triodes when viewed this way, and
some triodes are rather "peaky" like this one.
Shouldn't there be no peak if the tube was perfectly OK?
What would be the cause?
(BTW, data taken using $3.49 Cen-Tech DVOMs from Harbor Freight, one
used for each grid, and one for the plate via a switch)

OK; measured cathode-to-plate voltage in the GR Megohmmeter; is 38.7V.
Best i could do was to change my supply to 42.9V as choice of zeners
was rather limited.
Also measured grid-to-cathode voltage where the source was stiff WRT
a 10Meg DVM (Check: -2.138, Disch: -1.859, 100Kmeg: -1.860V).
Then i pulled the (almost unbalanced) tube from the Megohmmeter and
made measurements and saw a few surprises.
First, and most delightful, was that "peak" was almost totally gone.
Second, the grid current was a lot larger at -0.5V relative to 100V
plate.
Not too surprising was that the grid current was substantially lower
at "reasonable" biases; a lot less that i would have guessed.
The Red curve is for the grid in the hi-z arm, where the unknown
resistor goes.
Seems that there is very little difference in the grid currents,
which tends to imply that the "meter balance" pot setting may be rather
sensitive.
No adjustment pots are shown in the simplified schematic.
Suffice to say the FP controls "Check" and "set infinity" are between
the cathodes, and the thru-panel "meter balance" (unlabelled) adjustment
seems to vary the voltage on the reference and unknown resistors.
**
Now i do not know what the heck to look for in the tube data...