exhausted to a high point, and No. 2 generator was set to work. Hydrogen was passed several times at full pressure through the appa- ratus for one or two hours, and then exhausted to the stratification point. During these operations the platinum terminals of one of the vacuum tubes were heated to full redness, and the current was kept on both tubes for some hours to drive off occluded gases.
Finally the second generator was sealed off, and hydrogen used from the remaining generator. After much washing out with hydrogen at the ordinary pressure, exhaustion and re-filling were continued, and finally the reservoir K was filled, both taps, H and I, being closed.. The tubes were highly exhausted to the non-conducting point, and tap I opened and then closed, so as to introduce a little hydrogen. H was then opened and again closed, so as to equalise the pressure in I, and exhaustion proceeded to the stratification point. At first the strata were irregularly coloured with a suspicion of blue on one face,, but as the operations just described were continued, the blue faces dis- appeared, the stratifications assumed a pure pink hue, and showed the hydrogen spectrum alone ; no mercury was detected in any part of either tube.
From the first to the eighth filling the strata were pink with a trace of slaty blue colour on the faces next the negative pole. From the tenth filling the blue faces disappeared, and after the twentieth filling no trace of blue could be seen, and the spectrum of hydrogen alone was visible.
On examining the spectra of the stratified gas in the two tubes, each showed strongly the line spectrum of hydrogen ; but while the spec- trum in the platinum-poled tube showed pure red, blue, and green lines on a black ground, that in the aluminium-poled tube showed in addition the fainter hydrogen line spectrum in the yellow and orange. This result may be due to the greater surface exposed by the aluminium poles ; it Avas not further examined.
Having at last succeeded in getting hydrogen free from mercury,, experiments were instituted to verify the inference that the blue com- ponents of the blue and pink strata usually attributed to hydrogen were really due to the presence of a trace of mercury.
Origin of the Blue Component of Parti-coloured Stratifications.
I used an apparatus similar to the last, but with only one generator. If my idea was correct, that the mercury in the course of a few hours diffused into the hydrogen tube from the pump when it was not at work, there ought to be an access of blue faces to the pink buttons after the exhausted apparatus had been at rest. After filling with hydrogen and exhausting several times, a hydrogen vacuum was obtained showing no blue faces to the pink strata. The apparatus was
