6 



Proceedings of the lloyal IrUli Academy. 



second slide can be adjusted by screws at one end. A steel scale (B) 

 divided into millimeters, and read by a vernier, serves to identify the 

 object viewed in tlie microscope. The latter is at the left end of a 

 metal frame moving on a second slide (A^, at right angles to and of 

 similar construction to the first one ; it is also supj^liecl with clamp, 

 slow motion, and steel scale for identification. The viewing micro- 

 scope is only provided with two cross-wires. 29-5 cm. to the right 

 of it on the same frame, and therefore rigidly connected with it, 

 is the measuring micrometer microscope, through which is seen a 

 rectilinear glass scale, supported in a separate frame on the gun- 

 metal base of the instrument, and parallel to the A-slide. This 

 microscope is of the form usual on transit circles ; the screw carries 

 two close, parallel wires, and the drum of the screw is divided into a 

 hundred parts, tenths of which can be estimated. One revolution of 

 the screw is equal to one-fifth of a millimeter, or one-fifth of the 

 interval between two consecutive division lines on the scale ; the 

 number of turns is counted on a comb in the usual manner. 



AYlien the x^hotographic plate is placed so that north is at the top 

 or bottom, the A-scale will give differences of abscissae approximately, 

 and the glass scale the same accurately, while the B- scale gives 

 approximate differences of ordinates. The plate may be turned 90°, 

 and the latter determined accurately. My practice has been to 

 determine first all the .r's in zones of 5 mm. breadth, and then all the 

 y's in the same manner, adopting a star near the centre of the plate as 

 zero point, and measuring carefully its coordinate at the beginning and 

 end of each zone, to make sure that the heat of the observer's body did 

 not affect the results. Care M'as taken to prevent this by cardboard 

 screens ; and whenever an appreciable change was found, a gradual 

 alteration was assumed to have taken place in the distance between 

 the two microscopes, otherwise a simple mean of the two values was 

 adopted.^ 



The screw has only been used for less than tliree revolutions to 

 either side of the central notch of the comb in the microscope; and only 

 that part of the screw has therefore been investigated for periodic and 

 progressive errors. To find the periodic errors, the distance between 

 the parallel wires was measured by setting first odc wire and then the 

 other on a small dot on the glass scale, the head reading very nearly 

 O'-O, O'^-l, . . . 0'-9. Three complete sets were taken of the six revo- 



1 On three occasions (out of thirty- six) the difference amounted to as much as 

 0-004 mm. or 0"-66, while twenty-one differences were less than 0*002 mm. or 

 0"-33. 



