THE MEASUREMENT OF CURRENT 37 



contacts or by corrosion. Rosin should be used as a flux, as it is 

 difficult to remove the last traces of acid, which would seriously 

 corrode the delicate wires. For the same reason, so-called non- 

 corrosive soldering liquids should be avoided. 



The use of the taut suspension employed in the original 

 D'Arsonval instrument is not advisable unless there is a special 

 provision for balancing the coil, for it is difficult to attach the 

 suspension wires so that their axes pass through the center of 

 gravity of the coil when the suspension is drawn taut. If this 

 condition is not fulfilled, the center of gravity is coerced into 

 taking up an abnormal position and the weight of the coil will 

 cause a turning moment which will vary with the tightness of 

 the wire and the level of the instrument. For these reasons, in 

 sensitive instruments of the best design, the coil is allowed to hang 

 free and the electrical connection at the bottom of the coil is 

 made by a loose spiral which may have a very small torsional 

 rigidity. This procedure also decreases the stress in the upper 

 wire so that it may be made smaller. However, in special cases 

 where the instrument is to be subjected to great changes of level, 

 as on shipboard, the taut wire must be employed. In the Sulli- 

 van marine galvanometer and other similar instruments, the 

 center of gravity of the moving coil can be adjusted to its proper 

 position by bending bits of lead wire which project from the coil 

 frame or by adjusting two sets of screws which project at right 

 angles through and perpendicular to the shank supporting the 

 coil. 



Effect of Changes of Temperature. If the instrument is to be 

 used as an unshunted current galvanometer, a change of room 

 temperature will alter the calibration. Experiments show that 

 for phosphor-bronze strip the elasticity decreases about 0.05 per 

 cent, per degree rise of temperature. The strength of the magnet 

 also diminishes with an increase of temperature, the change being 

 0.01 or 0.02 per cent, per degree. It varies with different mag- 

 nets. The tendency of these effects is toward compensation, but 

 in general their relative magnitudes will not be such as to elimi- 

 nate the error. 



If the instrument is shunted, the multiplying power of the 

 shunt will depend, to a certain extent, upon temperature con- 

 ditions, for the galvanometer is wound with copper and the 



