514 



NA rURE 



\_MarcIi 27, 1884 



THE REVERSAL OF HALL'S PHENOMENON 

 TN a recent cnmniunication to the Pby>ical Society I mentioi.ed 

 among other thinj^s that I had succeeded in reversing the 

 direction of tlie Hall effect in iron. It wa?, hinvever, found to 

 be so exceedingly dififKult to keep the two points v\here the 

 galvanometer connections were made at the ;anie potential, even 

 for a few seconds, that the extent of the deflections due to the 

 Hall effect could only be roughly guessed at, and the experiment 

 was hardly a satisfactory one. I believe this inconvenience aruNe 

 from the fact that the iron, being a strongly magnetic metal, was 

 slightly displaced whenever the polarity of the electromagnet 

 was reversed, thus shifting the points of contact with the gal- 

 vanometer wires. I have .Mnce repeated the experiment with 

 gold, which turns out to be perfectly easy to work with, and 

 altogether more suitable for the purpose. The following is an 

 acc'.unt of four experiments : — 



Experiment I. — A piece of nearly pure gold foil 5 cm. long 

 and 3'5 cm. broad was cemented to a plate of glass and the 

 whole placed between the flat pole pieces of an electromagnet. 

 The middle points, A, u (see figure) of the longer sides of the foil 

 were connected to a galvanometer, G, and the middle points, 

 c, F, of the shorter sides to a battery. A current was passed 

 through the metal from left to right, and the electromagnet 



excited so that a south pole was beneath the glass and a north 

 pole above it. The galvanometer was immediately deflected, 

 indicating a current flowing in the direction bg A. If either the 

 polarity of the magnet or the direction of the current through the 

 foil was reversed, the transverse current was also reversed and 

 flow ed in the direction A G B. This is the ordinary "Hall effect," 

 and the direction of the transverse currents agrees with that 

 mentioned I y Mr. Hall for gold. The extent of the deflections 

 varied from abi.ut 50 to 70 scale divisions on each side of zero. 

 Similar but smaller deflections occurred when the galvanometer 

 was connected with points nearer to the middle of the plate. 



Experiment 2. — Two longitudinal slits, F, H, about \ mm. 

 wide, were then cut along the middle of the foil, leaving a 

 connection 4 mm. wide between the two halves of the sheet, 

 and the former experiment was repeated. The following 

 are the details ; and to under^tand them it must be remembered 

 that the galvanomeler is affected by two causes besides the trans- 

 verse current : (i) by the direct action of the electromagnet upon 

 the galvanometer needle, though 13 feet away from it ; {2) by a 

 small permanent current due to the fact that, however carefully 

 adjusted, A and B are never (or hardly ever) at exactly the same 

 potential. 



The image of the galvanometer wire was brought as nearly as 

 possible to zero of the scale before beginning the experiment, 

 and the connections were made so that a current in the direction 

 AG B caustd a deflection to the left(-), and a current in the 

 direction EGA caused a deflection to the right ( + ). 



Upper pole of magnet north : — 



Galvanometer key, K, raised, deflection + 25 divs.* 

 ,, ,, depressed, ,, + 102 divs.- 



Due solely to the action of the magtlel upon the galvanometer needle. 



^ Due partly to the action of the magnet on the galvanomettr needle, 

 partly to the permanent current above referred to, and partly to the 

 transverse current resulting from magnetisatioD. | 



Upper pole of magnet south : — 



Galvanometer key raied, deflection - 24 divs. 

 >, ,, depressed, ,, - 42 divs. 



Net deflections due to current (subtracting effect of the magnet 

 on the galvanometer needle) : — • 



Upper pole north (102 - 25 = ) + 77 divs. 

 ,, south (-42 + 24 = ) - iS divs. 



Sum of oppcsite deflections due to transverse current, (77 *• 

 lS = )95, o'" deflection on each side of zero =47'S divs. 



The slits therefore had the effect of reducing the amount of 

 the Hall deflections ; the direction was unaffected. 



Experiment 3. — The galvanometer contacts were now moved 

 from tlie edges to the pomts D, E, about 5 mm. from the middle 

 line, and the experiment was repeated with the following 

 result : — 



Ujiper pole of magnet north : — 



Key raised, deflection + 18 divs. 

 „ depressed, ,, +165 divs. 

 Upi-er pole south : — 



Key raised, dt flection - 35 divs. 

 ,, depressed, ,, + 180 divs. 

 Net deflections due to current : — 



U| per pole north (165 - 18 = ) + 147 divs. 

 ,, south (i8o + 35 = ) + 2i5 divs. 



Sum of deflections c ue to transverse current (215 - 147 = ) 68. 

 Uefieclion on each side of zero = 34 divs. 



Thus when the galvanometer cf ntatts were near the middle of 

 the plate the deflections were almost as great as when the gal- 

 vanometer was connected to the edges. But ihcy were in the 

 offosite direction, showing that the Hall effect was reversed. 

 Experiment 4. — A repetition of the last. 

 Upper pole i.orlh : — 



Key raised, deflection + 28 divs. 

 ,, depressed, ,, +170 divs. 

 Upper pole south : — 



Key raised, deflection - 24 divs. 

 ,, depressed, ,, +170 divs. 

 Net deflections due to current : — 



Upper pole north (170-28 = ) 132 divs. 

 ,, south (1704-24 = ) 194 divs. 



Sum of deflections due to transverse current, (194 - 132 = ) 62. 

 Deflection on each side of zero =31 divs. 

 These results, curious as they are, were of coutse not unex- 

 pected, the experiment having tetn in fact devised for tiie 

 purpo;e of testing in an absolutely conclusive manner the suffi- 

 ciency of the explanation of Hall's phenomenon by strains and 

 Peltier effects which I have recently proposed (see Nature, 

 p. 467). 



Supposing the magnet and the battery to be so arranged that 

 befoie the slits were made the points A and D were in stretched 

 districts, and B and E in compressed districts of the metallic 

 sheet, then the effect of cutting the slils will be practically to 

 divide the plate into two independent p'ates, each of which 

 undergoes strains similar to those originally existing in the whole. 

 A and B therefore will still be in regions which are respectively 

 stretched and compressed, while on the other hand the region in 

 which D is will row be comyressed, and that in which E is wdl 

 be stretched. Thus as regards the points D and E the result of 

 making the slits is to reverse the strain, and in consequence the 

 Peltier effects and the galvanometer deflections. If iVlr. Hall's 

 own theory were correct, the existence cf the slits shtukl make 

 no appreciable difference of any kind. That they should have 

 the effect of rnming the action of the magnet upon the current 

 is altogether inconceivable. Shelford Bidwell 



DR. FEUSSNER'S .\El-V POLARISING PRISM 

 TN a recent number of the Zeits<hrift jiir Insirunicnttnkundc 

 (iv. 42-50, February 1884), Dr. K. Feu-sner cf Karlsruhe 

 has given a detailed description of a polarising prism Luely 

 devised by him, which presents several points of novelty, and 

 for which certain advantages are claimed. The paper also con- 

 tains an account, although not an exh.auslive one, of the various 

 polarising prisms which have from time to time been constructed 

 by means of different combinations of Iceland spar. The litera- 

 ture of this subject is scattered and somewhat difficult of access. 



