384 ACTUAL WORKING IN ORIGINAL RESEARCH. 



phenomenon of electric currents produced by two un- 

 equally heated pieces of platinum in liquids which did not 

 corrode them, and employed an apparatus which was 

 rendered water-tight by means of washers of pure india- 

 rubber. After making many experiments, I found that, 

 although the indiarubber had no chemical action on 

 platinum, it chemically altered some of the liquids, and 

 thus affected the electric currents, though not to such 

 a degree as seriously to influence the results. I had, how- 

 ever, to abandon the use of an expensive apparatus and 

 devise and employ a different one, devoid of indiarubber 

 in contact with the liquids, and also to repeat with it 

 most of the former experiments, in order to compare the 

 new results with the old ones, and thus determine the 

 extent of the interference. Also, in modifying an expe- 

 riment, or the method of working, in a research, whether 

 the modification is intended for the purpose of saving 

 time or expense, removing an inconvenience, or for other 

 reasons, it is necessary to consider previously whether in 

 so doing we may not introduce an interfering condition. 



As neither our apparatus, materials, or modes of mani- 

 pulation are perfect, the results obtained by means of 

 experiments must always be contaminated with more or 

 less error, and therefore extremely exact results are usually 

 suspicious. 



