42 
DE. A. SCHUSTEE ON THE SPECTEA OF METALLOIDS. 
have been formed in the negative half to diffuse into the other half, and then make 
contact either the same way as before, or the opposite way, the spectrum of the 
negative pole will be seen throughout the tube at first, and only gradually disappear 
in the positive half. 
Experiment 4.—Make and break contact very rapidly, always changing the direction 
of the current; after a little time, when the effect of the first make has passed 
away, the permanent state will at once be established for each current in either 
direction, so that in this case the spectrum characteristic of the negative glow is 
always only seen surrounding the negative electrode ; but I believe that the glow 
itself in this case is not so much developed. 
Experiment 5.—After alternating the current quickly, as in Experiment 4, you wait 
a little while, and then make contact either way; the spectrum of the negative pole 
will be at first seen throughout the tube, and gradually die away in the positive half. 
These experiments succeed as well in nitrogen as in oxygen. Experiments 4 and 5 
weaken somewhat the argument. I have obtained, however, other results on suddenly 
reversing the current, which require further investigation. Without pretending to be 
able to give a complete explanation of the experiments, it seems to be proved by them 
that the spectrum of the negative pole may appear away from the pole, and must, 
therefore, be due to secondary causes. I have proved in another place that a mere 
difference in temperature cannot account for the production of the spectrum at the 
negative pole.'" I cannot help quoting, in this place, a passage taken out of 
Earaday’s ‘Experimental Researches.’ After having referred to an explanation 
which makes the so-called unipolarity in the conduction of soap and similar bodies 
depend on secondary chemical phenomena, which take place at the poles, he says :— 
“ Probably the effects which have been called effects of unipolarity, and the peculiar 
differences of the positive and negative surface when discharging into air, gases or 
other dielectrics, which have been already referred to, may have considerable relation 
to each other. ”+ 
3. Description of Apparatus. 
In an investigation of the spectra of pure gases it is essentially necessary to avoid all 
traces of carbon impurities. Indiarubber tubings, as well as any joint involving the 
introduction of grease, paraffin, &c., cannot be used. I do not say that such joints are 
always detrimental to the correct experiment, but if the question has to be decided 
whether a certain spectrum is due to an impurity or not, all excuse for the presence of 
the impurity must be avoided. These precautions are absolutely necessary if a pure 
spectrum of oxygen is desired. I have examined a great number of tubes supposed to 
be filled with pure oxygen by various makers, but I have only seen one tube which did 
not show the spectrum of carbonic oxide. This spectrum has been described by various 
* Proceedings of tbe Cambridge Pliil. Soc., vol. iii., p. 57. 
t ‘ Experimental Eesearcbes,’ Series XIII., 1636. 
