Hi Jack,
that's a constant force mechanism!
During his countless hours trying to get it working, i.e. to achieve a little better rate stability, he realized that this escapement was very sensitive to this problem and he decided to add a constant force. You may imagine what that means once the base design is already done ;-)
The original idea (it's an old dream to somehow achieve a less harsh impulsing, I'm lacking the correct english term, perhaps it could be called "impact free" impulse?) goes back to Benoit and Himmelheber - bot designed torubillon escapements where the impulse is supplied through the hairspring (i.e. not directly to the balance).
As nice as it works in terms of smoothness of impulse, it's extremely problematic and complex to work precisely.
The exact timing for unlocking/impulse is most crucial and that's one of the weak parts of those designs.
BTW, Rieflers version with balance wheel (the pictures of Paul Gerbers table clock ) is similar and I'd say belongs to the same group.
Christian Klings combined this basic design with his desmodromic escapement - but as said even investing far more time and energy as expected - it never worked reliably. But looking at the running escapement is incredibly mesmerizing and one easily forgets about such sterile things as "rate"
If' I'd have had to guess if it could work before I actually saw it in the flesh - I wouldn't have believed it could.
Best
Suitbert