288 PROCEEDINGS OF THE AMERICAN ACADEMY. 



wire leading, through the mercury cup, W, to a galvanometer. Con- 

 nector C" does the same for junctions 1', 4', 7', and 10'. The same 

 connectors can evidently be used to bring junctions 2, 5, 8, 11, and 2', 

 5', 8', 11' into action, or 3, 6, 9, 12, and 3', 6', 9', 12'. The central 

 junctions, 13 and 13', were used by themselves. 



The apparatus, beyond this switch-board, does not require detailed 

 description. The resistance coils used were reduced to legal ohms 

 by comparison with tliose of an Anthony standard (Queen & Co.). 

 Thermometers at two or three points along the circuit gave such 

 account of the temperature as was necessary. The galvanometer was 

 a Thomson astatic, with a concave mirror. The strength of current 

 required to produce a deflection of 1 cm. on the scale was about 

 12 X 10"^ amperes. 



The circuit containing the copper-German-silver junctions, ^i and j\^ 

 of Figure 1, is simpler and requires little further description. The 

 galvanometer of this circuit had a plane mirror, and was used with tele- 

 scoi^e and scale. Its sensitiveness to a given current was about the 

 same as that of the first galvanometer. The circuit included, at a 

 point corresponding to the switch-board of the other circuit, mercury 

 cups at which the connection could be broken. 



In addition to the mercury cups already mentioned, each circuit 

 included a mercury commutator, by means of which the course of the 

 current in the galvanometer could be reversed. These mercury cups 

 and commutators were a source of much disturbance, and toward the 

 last of the work solid metallic connections were in some cases substi- 

 tuted for them. The difficulty was thermo-electric. If two copper 

 wires run into the same mercury cup at different levels, or if the two 

 wires are of different size or diflferent external condition, there will 

 be at the copper-mercury junctions, unless the temperature of the 

 room is more than usually uniform and constant, thermo-electric 

 effects which cannot be neglected in delicate measurements of electric 

 current. There was a certain approach to uniformity in these dis- 

 turbances, which made it possible to eliminate them in great part 

 from the result by the device, already described, of running the 

 warmer stream of water alternately above and beneath the disk. 

 The full measure of this source of error was, however, not realized 

 until the last week or two of the work, and it may be that the dis- 

 cordance between the results of August 13 and 15 and those ob- 

 tained later was due to lack of care in protecting the copper-mercury 

 junctions from changes of temperature on the days mentioned. 



