The only thing you can't see clearly in these pictures is the all-important support ball, or "bille" as Allais calls it. This consisted of a commercial SKF ball bearing of 6.5 mm diameter, claimed to be "high precision". By modern standards this was not high precision at all; my metrological contacts tell me that the ball was likely about grade 100, i.e. accurate to 100 millionths of an inch. Allais used a new support ball for each fourteen minute swinging episode, in an attempt to average out the errors due to inaccuracies of the balls, of which he was well aware. The ball rolled upon the support plate labeled S in the fourth and last figure (about which Allais tells us nothing), and was held in a conical depression in the (non-labeled) cylindrico-conical mount shown as projecting downward from the bracket E. So the entire pendulum was supported upon this little ball. The idea (a brilliant one) was to eliminate sliding friction and to have no friction except rolling friction.