THE BALANCE. 



have been collated in the following table, air being supposed at 0, 

 and under the pressure of 760 mm. 



5 5 



D Weight Weight 



D in Brass. in Platinum. 



Platinum.. 21*30 0060. io~ 94. io~ 6 o 



Mercury J 3'59 0095.10" 59.10" -15.10" 



Copper 8-85 0146.10" 8.io~ -86.10" 



Brass 8-40 0154.10" o.io~ -94.10" 



Aluminum 2-73 0473.10" -319.10" -413.10" 



Water i-oo 1293.10- -1139.10- -1233.10" 



With brass weights, the term of correction in weighings with the 

 metals is far below the ten-thousandth, except for platinum on the one 

 hand and aluminum on the other. There is no occasion to take into 

 account changes arising from alterations in temperature, for they do 

 not affect the corrections themselves to the extent of thousandths. 



The corrections in the weighings are relatively much more 

 important when the body we have in view forms part of a much 

 heavier system, and the weight is to be obtained from them by 

 difference as in chemical analyses, for instance. 



658. MEASUREMENT OF ANGLES. The evaluation of an angle, 

 which is in reality an abstract number, amounts to the measurement 

 of a length, or more exactly to the comparison of two lengths. As 

 this problem is often met with, and in the most varied conditions, 

 it may be useful to examine its general features. 



In most cases the angle is measured by the displacement ot 

 telescope movable over a divided circle. The telescope has : ist, a 

 line of sight denned by a limb or by a cross wire telescope, which 

 is alternately brought in the direction of the two sides of the angle ; 

 2nd, an index, or a vernier, the displacement of which is observed 

 in reference to the division of the circle. 



The accuracy of the measurements depends on the precision with 

 which the line of sight is divided, and, on the other hand, on the 

 graduation of the circle. In a well constructed instrument, the 

 approximation of the sighting should be at least equal to that of 

 the graduation. If, for instance, the circle with the vernier measures 

 10 seconds, the telescope which serves for the sighting should not 

 have an error greater than 10 seconds. 



659. When we look through a telescope at a luminous point 

 without apparent magnitude standing out against a dark ground 



B 2 



