Valence voltage is that intrinsic voltage drop of a diode, or in this case, a transistor. It varies inversely with temperature, and is closely related to the reading you get on your diode scale. Put your lighter on under the DUT and the voltage drops. In this case the heat is applied from the outside. The case temperature will be higher than the junction temp.
If the heat is emitted from the inside, the junction is hotter then the case. This is from whence we can derive the real figures.
So the .645V you read across a diode is the valence voltage at that current and temp. When driving the base of a transistor it is ever so much more complex.
What's more, 0.140 might be a valid forward reading, and in some devices it's not wise to test them in reverse. Even the 2 volts reverse might damage a Ge device. Some may even read leaky when they're good. Germanium transistors are not all that great except for one thing. They have low loss at lower voltage levels, like in battery powered stuff. Silicon devices outperform them hands down until we get to that intrinsic voltage drop. They have made great strides in silicon, but if the same effort were put into Ge, or even glass Ovonics, what might the result be ?
[glass ovonics is an anamorphic semiconductor, the technology was seemingly dropped after it introduction in the 1970s, very important facets of this technology have found their way into the manufacture of solar cells]
IIRC, glass ovonics was polarity independant also, and at the time it was demonstrated (yes it was) engineers were still having alot of fun having, for the first time, complementary semiconductors.
For about a decade they did, but a nonpolarized semi, not NPN nor PNP I think would have really put a craw in them. Twenty years before it was all tubes, well mostly.
Everybody thought in a negative ground sense, one of the most flagrant dismissals came with the Sony EXR series of RPTVs. It uses a negative supply for H deflection and as far as I can discern, the yokes are DC coupled.