42 



NATURE 



[November 14, 1895 



Again, the thick coating of paraffin, and the solid core, which 

 are almost universally prevalent, increase the uncertainty, for if 

 any temperature change is taking place the lag is considerable. 



These sources of error were diminished as follows : the sides 

 and bottom of the box are the inner walls of a double copper 

 tank, holding about eight gallons of water maintained at a 

 temperature near 20° C. by a regulator. Over the top of the 

 box and tank is fixed a case similar to that of an ordinary 

 balance, the front glass of which is only raised when adjustment 

 of plugs and contact-maker is necessary, all connecting screws 

 being exterior to the case. The silk-covered coils (which are 

 double in all cases to reduce the effects of current heating) are 

 suspended from an ebonit« rack within the box. They received 



( 



[TiiTiiii|iiiiiiiii|iiiiiiiii|iiiiiiiii|iniiiiimMiiiiii|iiiniiii 



BRASS BAR 



Fig. 2. 



the thinnest possible coating of paraffin and are wound loosely 

 and openly, so as to be subjected to no strain, and also to permit 

 free circulation of air to all parts of the wires. An open 

 scale mercury thermometer, with a small bulb, is placed with its 

 bulb contiguous to one of the higher resistance coils. A series 

 of observations leads to the conclusion that, under these circum- 

 stances, the thermometer-reading sufficiently indicates the real 

 temperature of the coils, provided of course that no rapid 

 changes are taking place, and this conclusion is borne out by 

 the observations of Messrs. Heycock and Neville on a box 

 similarly protected by a balance case. 



The coils were annealed in paraffin at a temperature slightly 

 below that which would carbonise the silk covering. 



y///y//^ 



The box-unit is (approximately) TT77rth of a Board of Trade 

 ohm. The coils have the following nominal values in terms of 

 this unit. 



A 



640, 



B 

 320, 



[60, 80, 40, 



FI 

 100, 



and the bridge-wire has such a section that a change of i cm. 

 in the reading indicates (approximately) a change of i unit in/ 

 (in reality i cm. = -9957 box units), and the resistance of the 

 thermometers, with one exception, is such that a change of one 

 box unit corresponds to l° on the platinum scale. 



The whole length of the bridge-wire is 30 cm. ; thus any 

 resistance exceeding 40 units can be measured in three or four 



NO. 1359. VOL. 53] 



different ways by changing plugs and bridge-wire contact* 

 Thus the accuracy of the various corrections can at any time be 

 exposed to a severe test. 



Great attention was given to the drawing of the bridge-wire^ 

 for, although the effect of irregularities would be eliminated by 

 the subsequent calibration, it was desirable to make it as 

 uniform as possible. In this matter the Scientific Instrument 

 Company were very successful, for it was ultimately ascertained 

 that if the wire was assumed as uniform, the greatest resulting 

 error would not exceed 0'005 units. 



Before its final attachment to the box, the wire was hung from 

 a conductor, and had a small weight fastened to the lower end^ 

 which communicated with a cup of mercury ; it was then 

 raised to a bright red heat by means of an 

 electric current, after which the cells were 

 gradually switched off, so that the cooling 

 was slow. The annealing was thus very 

 perfect, and the wire on being released re- 

 mained quite straight. The coefficient of 

 expansion of this platinum-silver alloy lies 

 between that of steel and brass. A narrow 

 parallelogram was formed, whose longer 

 sides consisted of brass and steel bars- 

 respectively — the shorter sides of ebonite. 

 The steel and iron bars were connected at 

 their centres to the marble box top, and. 

 the wire placed between, and paralieh to, 

 them, its ends being fixed to the ebonite 

 cross-pieces and connected by flexible brass 

 strips to the remainder of the bridge. By 

 this arrangement the tension of the bridge- 

 wire is kept constant when the tempera- 

 ture of the box alters, and, at the 

 same time, the zero-point at the centre remains unchanged ia 

 position. 



A second platinum-silver w^re, permanently connected with 

 the galvanometer, lies alongside the true bridge-wire. The 

 vernier-slide carries a small cross-bar of the same wire placed 

 beneath and at right angles to the bridge and galvanometer 

 wires, and only just clearing them (Fig. 3). When the contact-^ 

 screw is either forced, or screwed down, both wires are pressed 

 on to the cross-piece by means of pads, and arrangements are 

 made to prevent any pressure being exerted which could injure 

 the bridge-wire. This method has several advantages, one of 

 which is that only similar metals are brought in contact, and 

 thus thermo- electric effects at this junction are avoided. 



The vernier reads directly (by means of 

 a microscope) to Jj^th m.m., thus yiiyth 

 m.m. can be estimated ; i.e. approximately 

 0001 box units, or o'ooooi ohms. 



Great difficulty has hitherto been experi* 

 enced in constructing a fine adjustment 

 for a bridge-wire contact. It must be of 

 such a nature that it will permit the free 

 movement by hand of the contact-maker 

 to any position. Again, if, owing to an 

 oversight the contEKt-maker is screwed 

 down, and any of the ordinary means of 

 fine adjustrtient ar« used, the bridge-wire 

 is subjected to a scraping action which 

 may affect its section. These difficulties 

 have been overcome by an ingenious 

 device designed for this apparatus by Mr. 

 Horace Darwin. 



Fig. 2 is a plan, and Fig. 3 a vertical 

 ' section of the contact-maker. 



ABA' (Fig. 2) is a brass framework which slides between the 

 steel and brass bars previously referred to. An inner block 

 F E E' H stands within the brass framework with a play-space at 

 its ends of about i cm. Springs at A and A' press the brass 

 frame against the steel bar, and springs at E, E' press the inner 

 block against the front brass bar. Thus the pressure of the 

 brass frame against the steel bar is the sum of the pressures of 

 the springs at A, A', E and E', whereas the pressure of the 

 inner block on the front bar at F H is the sum of E and E' 

 only ; if, therefore, the screw S is rotated the inner block alone 

 is moved. As the screw S recedes the inner block is made to 

 follow it by means of long spring indicated in the plan by the 

 dotted lines with arrow-heads. If by inadvertence S is turned 

 when the bridge-wire has not been released by the screw C 



