Would like some help Reverse Engineering 2 parts of a machine made in 1905

From what I can tell reading the patent, the groove profile at its middle should have a radius equal to that of the finished ball surface. And the flange surface profiles either side should be tangent to the radius and have an included angle that will shape the preform into a rough sphere that will gradually feed into the radius profile surface area of the pulleys. The surface speed differential between the two pulleys will cause the ball to assume a uniform spherical shape when rolled in this circular space. And finally one of the pulley shafts is shifted slightly from the forming position, allowing the finished ball to squeeze thru.

You will need to adjust the pulley diameters to fit the existing machine, and they will be different for each ball size. As noted in the patent, the pulley flange OD's should be as large as practical without contacting each other during operation. You will likely need to run and measure a few samples to get the travel stop settings right. Also, since this machine only performs a rolling process, your material preforms will need to be fairly close to spherical to start with, and they should have an accurate/consistent volume of material.

The grove profile should be slightly larger than the desired ball radius. This machine can only make spherical whatever lump is fed into it. It cannot add or subtract material so the actual ball diameter is determined by the lump size. When the lump becomes hard enough to separate the spring loaded grooved roll, it passes through. It works like rolling a piece of putty between your hands to form a ball.