Derick Fay wrote
if I twist my torso to increase propulsive force by
an average .5%, this may be offset by a 2% average
loss on the d/p from being in a less stable position . . .
or the figures might be reversed. We just don't know.

That's why it's important to delay the twist of the torso until _after_ the
main part of the pole-push motion -- in both V1 (poling-side only) and V2.

So even with no force numbers, it's a reasonable guess that the interference
with the effectiveness of the pole-push is small -- though of course like
Derick says, it's not proof.

Then I look at several elite racer videos with single-frame-step, and I see
precise confirmation of that guess: The racers delay the twist of their
torso in both V1 (poling-side only) and V2, but on the non-poling side of
their V1, the elite racers start the torso twist immediately (because
there's no pole-push to interfere with).

Next I try it myself on rollerskis, and on snow, and the delayed torso-twist
move just _feels_ stronger than "quiet upper body". After that, if I still
needed more proof, I would do measured personal time trials (not for this
move, but I did some for the "forward step").

In the absence of these kinds of measurements, isn't
on-snow performance a better means to evaluate
technique recommendations than theoretical deduction?

Sounds good to me.

So where has anyone displayed any sort of controlled time-trial measurements
to justify the "quiet upper body" religion that seems to be sweeping much of
American coaching right now?

Now I'll offer some controlled time-trial experiments on "quiet upper body":
the winners of all the World Cup skate races. The fastest skater on the US
National team in actual World Cup races. The QUB experiment fails on every
criterion.

Using physics is not about relying solely on "theoretical deduction".

It seems to me that any discussion of physics without
some measurement of the magnitude of the forces
involved isn't too helpful.

I put some estimates of force magnitudes up on the web at
http://roberts-1.com/xcski/skate/power_model
Since then I've thought of some improvements to the model, and perhaps a
significant correction -- just haven't gotten around to it. I'd love to see
someone publish something better, but it's a start.

Ken