478 PROCEEDINGS OF THE AMERICAN ACADEMY, 



important advantages of Nernst's apparatus. The circumstance of being 

 able to use unplatinized electrodes relieved me of what would otherwise 

 have been an unavoidable source of contamination. The electrodes 

 were fixed at adjustable distances apart upon a frame of hard rubber, 

 which, when set over the thimble beaker, suspended the electrodes in the 

 contained water. Between them, but a little to one side, was suspended 

 a mercurial thermometer with bulb immersed. This arrangement of 

 test-cell is not recommended as the best that could be constructed to 

 serve tlie required purpose. It is described here principally to give 

 information of the conditions under which the experiment was performed. 

 It results from the construction that glass (of beaker and thermometer), 

 platinum, and air are the only objects in contact with the contained 

 liquid. If necessary the bulb of the thermometer and the inside of the 

 beaker could be coated with paraffine. To test the efficiency of the 

 conditions, I made the following control experiment. The test-cell was 

 set up with a content of 3 cc. of very pure water. Practically simulta- 

 neous readings of time, temperature, and the length (Z) of the measuring 

 column of resistance-liquid were made and recorded. The test-cell was 

 connected in parallel with the measuring tube ; it must be remembered 

 that in this mode of connection the length {L) increases with the conduc- 

 tivity of the liquid in the test-cell and vice versa. The numerical value 

 of the conductivity is not important to my argument. With the mirror 

 scale and the magnifier with cross-threails that were in use when this 

 experiment was made it was possible to read one twentieth of a scale- 

 division, cue tenth of a scale-division being a change in length of signifi- 

 cance, the scale-division being 1.2 mm. The position of the electrode was 

 read directly. The dial attachment (Nernst, '94, and Maltby, '95) for 

 very much smaller readings was not; used in this experiment. 



Under the control conditions above described the following record was 

 secured. 



Time, 9.25 9.30 Variation = -i-60 minutes. 



Temp., 20°. 80 20°. 90 Variation = — 0°.l C. 

 L., 10.5 10.5 Variation — divisions. 



The result shows satisfactorily that the apparatus itself (that is, the 

 test-cell) is not a source of contamination to the contained medium, when 

 iu use for a period of one hour. 



