I had seen beautiful bell shaped distribution of torque (around 0) from the Heisenberg’s lab. We thus checked the data we have (the 6 minutes data we produced with the same flies for the torque meter and compensator. Data produced on the same day, or later (when sathish was mastering the preparation a bit better).
Here is the distributions:

Our distribution are close enough to the bell shape obtain by the Heisenberg’s group. The wing beat analyser seem to lead to different torque calculation, though.
PS: no difference seen in the frequency of spikes on the other hand.