1871.] 



Action of Metals and Liquids. 



337 



of sulphate of magnesia, and also of hydrochloric acid and of chloride of 

 sodium, probably because two causes operated, viz. increased strength of 

 solution and diminished conduction-resistance ; in a very few cases, how- 

 ever, the opposite result took place, as with solutions of chloride and 

 nitrate of strontium. 



Inversions of the direction of the deflection by difference of strength of 

 the liquid occurred with solutions of chloride of copper, sulphate of zinc, 

 nitrate, iodide, and sulphate of sodium, bromide and iodide of potassium. 



Irregularities of the amount of deflection were very apt to take place 

 with liquids which gave strong deflections, or which acted much upon the 

 copper plates (for instance, nitric acid), especially if bubbles of air remained 

 under the plates, or the dishes were wetted on their side above the liquid 

 by the solution. 



In certain acid liquids, viz. nitric, chloric, hydrobromic, hydrofluosilicic, 

 and sulphuric acids, the hot copper was strongly negative (notwithstanding 

 the chemical action upon it was distinct, and 'in some cases even strong) ; 

 this is similar to the electrical behaviour of platinum in such liquids, and 

 may be attributed either to the more direct influence of the heat alone 

 (such as occurs with platinum plates), or to a different influence of the 

 chemical action produced by the heat. Both these causes probably ope- 

 rate in such cases. 



It is probable that in all cases where the hot copper was positive in 

 liquids of strongly acid reaction, the positive condition was due to chemical 

 action alone. 



With some liquids, especially with solutions of hydrocyanic, boracic, 

 acetic, tartaric, and citric acids, the deflections were very feeble, and the 

 chemical action on the plates not perceptible ; whilst with others, such as 

 nitric and chloric acids, solutions of the chlorides of strontium, sodium, 

 potassium, and ammonium, and of carbonate, acid carbonate, and cyanide 

 of potassium, the deflections were considerable, and the chemical action 

 distinct, and in some cases strong. In none of the liquids (except hydro- 

 bromic and chromic acids) did the hot plate appear to be less stained or 

 corroded than the cold one ; probably in all cases it was the most corroded, 

 although in some cases the corrosion was not perceptible. 



The amount of deflection was not always proportionate to the amount of 

 chemical action ; for instance, with solutions of chloride of copper and 

 iodate of potassium there was considerable corrosion, but only feeble cur- 

 rents, probably because the plates became covered with a badly conducting 

 film, whilst with hydrochloric acid, chloride of cobalt, chloride of manga- 

 nese, and nitrate of potassium the reverse occurred. 



I consider the currents in all these experiments of difference of tempera- 

 ture to be due either, 1st, to the direct influence of heat, the effect of which 

 is to make the hot copper negative in acid liquids and positive in alkaline 

 ones (see Phil. Mag. 1857, vol. xiii. p. 1); 2nd, to chemical action, which 

 sometimes overpowers the direct influence of heat and reverses the effect ; 



