948 



RECORD OF CURRENT RESEARCHES RELATING TO 



to zero, and tlie setting is made on line 2 by means of tlio screw c. 

 In moving over the space from line 2 to line 3 with the screw h, 

 it will be seen that the same part of the screw is used as in going from 

 line 1 to line 2. Hence the comparison of these two spaces is inde- 

 pendent of the errors of the comparing screw. 



Fig. 1. 



The number of spaces whicli can be compared in this way is only 

 limited by the length of the screw c, and the length of the opening 

 through the bed-plate. 



Again, suppose we measure the spaces 1, 2, 3, 4, . . . 100 by a 

 continuous forward motion of the screw. Such measures will in- 

 volve all the errors of the screw itself. But if after the measures are 

 made we set the screw h back at zero, turn the ruled plate around 

 180°, and set on line 100 with screw c, the continuous forward motion 

 of the screw b from line 100 to line 1 will be over the same part of 

 the screw as from line 1 to line 100. In the first case the screw 

 measures the accumulated errors of the ruled plate from line 1 to 

 any point up to line 100, but such measures involve the errors of the 

 comparing screw. In the second case the accumulated errors are 



