EXPERIMENT STATION BULLETINS. 577 



ated. The couiiections on the double contact switch should be made so as 

 to close the bridge circuit first and then the galvanometer circuit. A 

 bottle of damping fluid is supplied with the instrument. Enough of this 

 Quid is placed in the glass receptacle to cover the mica vane. The instru- 

 ment is thus made aperiodic coming back to zero rapidly without over 

 shooting. 



BRIDGE. 



The bridge, Fig. 11, is also made by the Leeds & Northrup Company. 

 It is the latest improvement of the roller type. Kohlrausch (14) early 

 showed that if an accuracy of .05 of 1% is desired the bridge wire would 

 have to be extended. Two coils, non-inductively wound, each of which has 

 a resistance of 4i/^ times that of the bridge wire itself, were placed at 

 opposite ends of the wire. This constitutes the "extended bridge." The 

 coils were so mounted on the bridge that by the removal of a couple of 

 plugs the extension coils could be connected in. The rheostat coils usu- 

 ally found on this type of bridge were disconnected to avoid any error 

 from capacity and inductance. The bridge has ten turns of wire upon its 

 drum with a resistance of 7.157 ohms. Each extension has a value of 32.2 

 ohms thus making the resistance of the bridge wire and its extensions 

 71.557 ohms. A wire of low resistance and larger cross section is prefer- 

 able to one that is very slight and slender because its calibration will re- 

 main more nearly constant. In addition it will stand wear and heavy 

 treatment better. 



There are a thousand scale divisions on the bridge and each one of these 

 is divided into halves, making a scale of two thousand parts. Each of the 

 smallest scale divisions is 3 mm. wide, and the bridge reading can be 

 made to 1/5 part of this division without much difficulty. The error from 

 a bridge reading will not be more than .002 of 1%. 



The alternating current was first passed through a variable resistance 

 in series with the bridge and then connected to the slider. The resistance 

 was introduced in order to vary the current, for when the more concen- 

 trated solutions are used much less current is needed. If the full cur- 

 rent was passed through, the heating of the electrolyte would follow, and 

 result in a false reading on the bridge. The resistance is connected on one 

 arm of the bridge and the electrolytic cell in the other. The swinging 

 coils are placed across the bridge. 



RESISTANCES. 



The known resistances must of necessity be the best obtainable. These 

 are the new Curtis Coils (Fig. 12). Their use was suggested to us after 

 reading the article by Dr. Curtis (4) in a publication of the U. S. Bureau 

 of Standards. According to a new method of winding the self -inductance 

 and capacity of the coils are anulled. By request the Leeds & Northrup 

 Company made for us a number of such coils of different values and as- 

 sembled them in a box. Two of tliese boxes have been used in the ex- 

 periments described in this paper. In assembling the coils in box form, 

 it should be remembered that it is necessary to keep them sufficiently 

 far apart to avoid any error from capacitj^ The newer "dial decade type" 

 is more convenient and has a range of from 1 to over 100,000 ohms. A 

 variable condenser (Fig. 12) must be introduced across the resistance to 



73 



