40 REPORT — 1861, » 



repeatiug tliis experiment, due to Seebeck and Magnus, I found that if tlie two ends 

 of two such copper -v^ares (being equally oxidized and annealed) were looped toge- 

 ther and held tightly in contact, little or no current could be observed when one of 

 the loops was heated in the flame ; but when the hot and cold loops were separated, 

 I observed a momentary current in the same direction as that produced when the 

 hot and cold wire were s\xA(\.erAy joined — i. e. from heat to cold across the joint. 



This fact, accidentally discovered, excited my attention, and led me to consider 

 what the acts of making and breaking contact could possibly have in conunon. I re- 

 flected that when two wires are approachmg or receding, they equally pass through 

 points at every possible distance (within limits) one from the other. Thus I thought 

 that the relative distance between the two wires might be the pecidiarity which, 

 being common to the two acts, might produce similar effects in each case. 1 there- 

 fore tried the effect of a loose contact between the two wires, resting the one wire 

 veiy lightly on the other, instead of pressing or pulling the two together ; a per- 

 manent cm'rent was at once produced, so strong as to hold the deflecting magnet of 

 the galvanometer against its limiting steps. I then introduced resistance cods into the 

 circuit for the pimpose of reducing the deflection, but to my sm-jirise it was not imtil 

 I had added a resistance equal to that of 2000 miles of the Red Sea cable, or about 

 1000 miles of the common No. 16 copper, that I reduced the deflections within the 

 range of my galvanometer. 



The current could be maintained through this resistance for twenty minutes at 

 a time — not perfectly constant indeed, but not wavering more than was inevitable 

 from the vai-jing pressure given by the hand to the two wires. The current was 

 etrongest when one end of the wires was white-hot, the other being dark red. 



I varied the experiment in many ways, using different galvanometers and differ- 

 ent copper wires, but always -svith one result. A tight contact gave a barely sen- 

 sible ciuTcnt ; a loose contact gave a current which could be maintained permanently 

 equal to that which would be produced through a similar resistance by the eighth 

 or tenth part of a Daniell's coil, — a strength suflicient to signal through a cable to 

 America, if ever one be laid. 



I next tried the same experiment with iron wires. Analogous residts were ob- 

 tained, but with one remarkable difference, namely, that the direction of the cm- 

 rent was from cold to hot across the joint, instead of from hot to cold as in copper: 

 moreover, a very sensible current was always observed in iron even when the two 

 loops were firmly held together; it seems possible that this effect is only a residue 

 of the effect caused by a loose contact, the hard oxide of iron precluding a perfect 

 metallic contact between the loops. The effect is increased at least fivefold when 

 a loose contact is made. 



The maximmn electromotive force to be obtained from iron is about one-twen- 

 tieth that given by copper, and acts in the opposite direction. 



Platinmn gives no current ■^'ith tight contacts ; with loose contacts a weak cur- 

 rent ffows in the same du-ection as that given by copper. I must here warn any 

 one disposed to repeat these experiments, that the resistance of the loose contact is 

 itself considerable; and if the whole circuit, including the galvanometer, be of small 

 resistance, the strongest deflection will be obtained with comparatively tight con- 

 tacts; for although the electromotive force is increased by loosening the contact, 

 the total resistance of the current may be increased in a still higher proportion, 

 and the strength of the ciuTent will then diminish. This effect is exactly analogous 

 to the well-kno-wn fact in voltaic electiicity ; for by the addition of small cells in 

 series to a battery with large sinfoce, the strength of the cun-ent may be reduced if 

 the total resistance of the circuit be small, but will be increased if the total resist- 

 ance be large. Thus the eflects of loose contacts are best seen on a sensitive gal- 

 vanometer with a large resistance in circuit. 



These phenomena may apparently be due to a thermo-electric absorption of heat 

 at the joint, or to a chemical efiect in one of the wires, the air or oxide acting as 

 an electrolyte. The opposite direction of the current in iron and copper, however, 

 gives a reason for believing that chemical action is not the cause of the current. 

 The decided effect obtained with platinum is another argimient for this belief It 

 is moreover well established that any variation in the molecular structiu-e of a 

 metal causes one part to become thermo-electrically positive or negative with respect 

 to the other ; thus thermo-electric couples can be made of hai'd and soft wire of one 



