THE ball -turning devic e illustrated in the accompanying drawings was designed to meet a special order for a large number of ball connections. A number of methods of doing the work were tried, but in every case the balls were not peI1fectly spherical and it was necessary to lap them in their seats. The fnal solution of the difculty was that shown in the drawings. The ball connection A (Fig. 1) was caught in the lathe chuck. The compound rest of the lathe was removed and a worm wheel SUbstituted. This had a pin in the center, which was fitted to the counterbore of the cross slide. The worm wheel was also supported by means of poles which fitted in a circular T-slot, though not so tightly as to prevent the wheel from revolving upon the center pivot. A tool post was mounted on the worm wheel, and .n this the tool was carried. The worm wheel which carried the tool was turned by a worm mounted on a shaft D, which was rotated by means of a belt 0 driven from the cone pulley of the lathe. The shaft D was connected to the pulley shaft by means of universal jOints, as shown clearly in Figs. 1 and 2. After a cut was taken the tool was advanced to a new position by a turn of the handle E. The balls turned in this manner were found to be perfectly true.