OF ARTS AND SCIENCES. 295 



following way. The bed of the comparator is made of cast-iron, 

 and ia sixty inches long by fourteen inches wide. It has an extreme 

 depth of two inches. In the centre, V-shaped ways run the entire 

 length of the bed, upon which) a plate carrying the microscopes, slides. 

 The bed has at one end the means of bringing it into a horizontal 

 plane, and at intermediate points screws for taking up the flexure. 

 For this purpose, free vertical bolts, pressed upwards by means of 

 levers and controlled by weights, are without doubt better than 

 screws. 



It is now necessary to provide for the movement of the micro- 

 scope slide in a horizontal plane. This is accomplished in the fol- 

 lowing way. A shallow dish of mercury is placed upon the bed 

 of the comparator, extending along its entire length. An arm pro- 

 jects from the microscope plate, to which is attached a plate sliding 

 between guides, and carried by a micrometer-screw. To the lower 

 part of this slide a platinum point is attached. One wire of a battery 

 having a single cell is attached to the platinum. Another is placed in 

 contact with the mercury. A sounder is placed in the circuit. The 

 microscope plate is moved to one end, and the platinum point is 

 brought into contact with the mercury, the contact being indicated by 

 the " click" of the magnet. The slide is then moved to the other end, 

 which is elevated or depressed by means of the adjusting screws, till 

 the platinum point again makes contact with the mercury. After one, 

 or at the most two trials, it will be found that the two ends of the bed- 

 plate are in the same horizontal plane. The microscope plate is now 

 set at the middle of the comparator, and- the amount of the flexure is 

 measured with the micrometer-screw. After about one third of the 

 measured amount has been taken up by means of the flexure screws, 

 the entire operation is repeated. In this way I find that the micro- 

 scope plate can be made to move sensibly in a true plane. In prac- 

 tice, it is found that almost equally good results can be obtained by 

 directly observing the surface of the mercury, with an objective of 

 pretty high power, using the interior illuminator. The surface of the 

 mercury admits of nearly as sharp a focus as the surface of a metal 

 bar. Good results have also been obtained by dropping fine threads 

 of spun glass upon the surface of the mercury. This method is rather 

 more convenient than the method by contacts, but the latter admits 

 of somewhat greater precision. From a limited number of trials, I 

 conclude that contacts can be made with a probable error of a single 

 contact not exceeding .00002 inch. 



Knowing that the microscope plate moves in a true plane, the sur- 



