248 KANSAS UNIVERSITY SCIENCE BULLETIN. 



thermometer. Both walls of the vacuum jacket are silvered from 

 the top to the bottom of the inner tube, to prevent radiation, 

 and so the usual platinum cylinder about the thermometer 

 bulb is unnecessary ; this avoids differences in concentration in 

 the tube, due to water returned from the condenser. There is 

 a question, however, whether the lack of the platinum cylinder 

 does not cause a slight variable depression of the boiling-point, 

 due to the cooler water from the condenser. Heat is supplied 

 by passing an electric current through coil B, which is welded 

 to heavy platinum terminals, HH, KK, passing through both 

 walls of the Dewar tube. The resistance of the coil at 100'" C. 

 is 1.39 ohms. 



It is a very simple piece of apparatus to handle, as it is all 

 welded into one piece, with the exception of the outer wall of 

 the condenser; this is removable, as in the ordinary Liebig 

 form. An ordinary Beckmann thermometer, graduated in hun- 

 dredths, made by F. 0. R. Goetze, in Leipzig, was used. When 

 in use, the heating coil was covered to a depth of 2.5 cm. with 

 garnets, and the lower end of the thermometer bulb was put 

 about 1 cm. above the garnets. It was found that a current of 

 3.8 amperes through the cell would cause 10 c.c. of water, just 

 enough to cover the thermometer bulb, to boil ; and that any 

 current from 4.4 to 4.8 amperes (27 to 32 watts) would cause 

 a very steady boiling, with any amount of water in the tube, 

 from 10 c.c. up to 30 c.c. Any greater volume up to 45 c.c, 

 enough to make the water bubble to the top of the Dewar tube, 

 gave a rather variable boiling-point, probably due to the cool- 

 ing effect of the upper part of the tube which was in contact 

 with the outer air. The current was furnished by a storage 

 battery, was measured with a Weston ammeter reading to 

 tenths, and was controlled by a rheostat. 



In the course of a set of readings, the thermometer was read 

 twice with an interval of five minutes, with the same amount 

 of solute and of the solvent ; and the average of these two read- 

 ings was taken as the correct reading for that condition. When 

 the condition was changed by adding more of either the solute 

 or of the solvent, an interval of ten minutes was allowed to let 

 the thermometer become stationary. It was found that the two 

 readings for the same condition rarely differed more than from 

 .001° to .004"^, unless there were more than 30 c.c. of liquid in 



