80 H. J. Grayson : 



of action and direct adjustment upon the divided circle with the- 

 aid of the microscopes was made by hand. Once this adjustment 

 had been made for eacli tooth, the whole system was clamped and 

 remained rigid during the passage of the grinding disc across the 

 tooth face. The precaution to complete each circuit of the wheel at 

 one operation without a break was always observed. Further-^ 

 more, every fresh traverse was commenced a quadrant to the rear of 

 the one preceding it. Adequate provision was also made to eliminate- 

 so far as was possible the chances of error due to fatigue or interrup- 

 tion; tlie latter being by far the most difficult to provide against. 

 It now remains to indicate in general terms the probable degree- 

 of accuracy which was attained by the foregoing method of cutting- 

 a ratchet wheel. Prolonged tests with a view to determine the- 

 nature and extent of the inaccuracies upon the circuit of the ratchet 

 were made. Necessarily errors are present even in the most care- 

 fully executed work of the character just described. It was hoped 

 that, assuming necessary precautions had l^een taken throughout,, 

 the degree of accuracy present in the divided circle Avhich was copied. 

 could at last be closely approximated, especially as the original 

 errors of this circle, which had already been proved to be small, 

 were bisected by the use of two microscopes in the adjustment made 

 for every tooth cut. 



Among the various tests made to determine the extent of remain- 

 ing inaccuracies, was one Avhich involved the use of a test indicator^ 

 not unlike the one already described and used for testing the screw. 

 In this case, however, it was made and fixed so that it engaged the 

 working face of a tooth similarly to the way in which the pawl 

 engaged it. It was so fixed that, as the ratchet wheel was moved on 

 the ruling engine by the pawl, it engaged first the teeth distant LSQo 

 from the pawl, and readings were taken. Then it was placed 

 at other distances as requii-ed in tlie usual nietliods of cali- 

 brating divided circles. This test was carried out upon the ruling^ 

 machine under conditions, so far as individual teeth were con- 

 cerned, analogous to those in operation when ruling. The magni- 

 fication used to detect movement of the index in the field of the- 

 microscope was such that a movement indicating an error of one- 

 thousandth part of a tooth interval could be read with ease. The 

 result deduced from a sei'ies of tests by the above method indicated 

 that the maximum error of any single tooth throughout the circum- 

 ference of the ratcliet was less than one-600th part of a tooth space 

 and no periodic error was perceptible. Unfortunately the limits. 



