the Thermo- Electric Properties of Liquids. 



515 



cooled and filtered; the presence of air in the liquid was, however, 

 found to interfere but little with the chief results, except with certain 

 solutions, or unless the air was large in amount. It is necessary that 

 the liquids be clear, for if a solid deposit settles upon the lower plate, 

 it interferes with the current. 



In using the apparatus it was essential to screw it together very 

 tightly before each experiment, otherwise it became somewhat 

 loosened by the influence of the heat upon the upper washers and the 

 screws, and the currents were thereby greatly increased in magnitude. 

 The quantity of water put into the cistern above the upper plate before 

 each experiment was two ounces, and the current of steam passed into 

 it raised its temperature from 60° to 210° F. in about seven or eight 

 minutes. It was necessary to take the apparatus asunder, and wash the 

 plates and washers thoroughly before each experiment, and to remove 

 from them every particle of extraneous substance. In many cases the 

 gold or platinum plates (but never the palladium ones, because they 

 oxidize) were heated to redness before an experiment. The galvano- 

 meter employed was in nearly all cases an astatic one, with a coil 

 offering a resistance of 100 ohms, and was found to be more suitable 

 than others which offered less resistance. 



I examined, by means of this apparatus, a large number of conduct- 

 ing solutions, in order to include conspicuous instances of the pheno- 

 mena and exceptional cases, as well as to more completely ascertain 

 any sources of interference, and the proper conditions for obtaining 

 the most reliable results. Metals or liquids which chemically affected 

 each other were usually avoided, and base metals, and even silver, 

 were rarely used. Plates of palladium were found to be much more fre- 

 quently corroded than those of platinum or gold, and were therefore 

 much less suitable for these experiments; they were thus acted upon 

 in hot or cold ones of chloride of chromium or perchloride of iron, 

 in hot solutions of either chloride of manganese, acidulated ones of 

 nitrate or chloride of copper, and in one of iodate of potassium. 

 Those of copper were corroded in a hot solution of carbonate of sodium. 

 With different liquids, also, different circumstances occurred, which 

 interfered with the currents ; in some cases the liquid corroded both 

 plates, as with palladium in chloride of chromium or perchloride of 

 iron ; in others it corroded the hot plate only. In some instances the 

 liquids were chemically changed by contact either with the hot plate 

 or the washers, and deposited oxide or basic salt upon them ; for 

 instance, perchloride of iron in contact with hot palladium, platinum, 

 or gold ; or permanganate of potassium in contact with hot platinum 

 or with the washers. With strongly alkaline solutions, the prepared 

 washers, and even those formed of pure rubber only, were acted upon, 

 the upper stratum of liquid became slightly yellow, and therefore 

 altered in chemical composition, and the currents were greatly altered 



