668 



SCIENCE 



[N. S. Vol. XXXIII. No. 852 



in the circuit an ammeter, a voltmeter and a 

 switch as shown in the figure. 



The condensing chamber HK may have a 

 variety of forms — the one sketched possibly 

 serves the purpose best. It will be noticed 

 that the condensed mercury is, by the inclina- 

 tion of the tube H, diverted and leaves the 

 still through a capillary delivery tube M^E^ 

 bent into the form of an S at its lower end. 

 The action of the mercury dropping into this 

 tube is that of a continuous mercury pump. 

 The nipple N for initially exhausting the sys- 

 tem is fused, for convenience, to the upper 

 end of this delivery tube. M^, M^ M„ M, are 

 mercury seals. 



The operation of the still is very simple and 

 when once under way needs but little atten- 

 tion. Fill the vessels E^ and E^ with the mer- 

 cury to be distilled and allow the m.outh of 

 the delivery tube to dip into a small beaker 

 of clean mercury. Now start the exhaust 

 pump (an efficient mechanical pump will 

 answer) and adjust the position of the vessels 

 E^ and E, until the electrodes A and B rise 

 to the heights indicated in the figure. The 

 mercury in the delivery tube should now stand 

 at some point p 5 or 6 cm. below the nipple N. 

 To start the are it is only necessary to lift the 

 vessel E^ slightly and allow a momentary 

 stream of mercury to flow down the tube AB. 

 If the vacuum is right the arc should start 

 instantly. Adjust the current to the proper 

 value. After fifteen or twenty minutes the 

 valve V may be closed, provided the various 

 mercury seals do not leak. The supply mer- 

 cury is drawn mostly from the vessel E^ and 

 it therefore should be of large surface and 

 rather shallow. The vessel E^ is provided with 

 an overflow. 



The rate of distillation depends upon the 

 size of the apparatus and the strength of cur- 

 rent employed. In the stills as flrst con- 

 structed the diameter of the tube AB was 

 about 20 mm. This gave for a current of 

 4 amperes approximately one pound per hour. 

 The corresponding fall of potential across the 

 terminals was 21 volts. Later the diameter of 

 AB was increased to 40 mm. and its length 

 to 25 cm. The size of the condensing cham- 



ber HK was proportionately larger. This still 

 gave for a current of 10 amperes approxi- 

 mately 2 pounds per hour. The potential dif- 

 ference across the terminals was 23 volts. 



To test the purity of the distillate zinc 

 amalgams were used. The test for zinc was 

 made by the electromotive-force method de- 

 scribed by Hulett. One millimeter deflection 

 of the galvanometer corresponded to approxi- 

 mately .0005 volt. The results are given in 

 the following table : 



Trom numbers 2a and 100 we see that a 

 deflection of 1 mm. corresponded to the pres- 

 ence of zinc in the ratio of 1 : 1,500,000. 

 Samples a, h, c were from an amalgam that 

 was, comparatively speaking, ' very impure to 

 zinc, yet the distillates condensed in three 

 separate condensing chambers showed prac- 

 tically no trace of zinc. The standard against 

 which the above was balanced in the test cell 

 was newly purchased commercial double dis- 

 tilled mercury and was in addition carefully 

 and repeatedly purified by the " wet " method. 

 Numerous additional tests under various con- 

 ditions have been made and all show that the 

 still is capable of giving a distillate of the 

 highest purity. In fact, very pure mercury 

 may be obtained by a single distillation even 

 though the original is very impure. As an 

 example may be given a test in which the 

 original mercury was excessively impure to 

 zinc. After distilling ofE six or eight pounds 

 the current was broken and the apparatus let 

 stand over night. In the morning the anode 

 was completely covered with a layer of zinc an 

 eighth of an inch thick that had crystallized 

 out on cooling. The distillate showed scarcely 

 a trace of zinc. In this instance the arc was 

 maintained by a small current density and 

 consequently the average temperature' in the 



•Knipp, Phys. Rev., Vol. XXXI., August, 1910. 



