ELECTIIIC DISCHARGES IN GASES AT LOW PRESSURES. 
427 
potentials first appears in the corresponding curves of Earhart. These deduced 
curves then indicate rapidly increasing spark potentials down to the 5-micra line. It 
will be seen, too, that this feature of the curves extends over a range of spark lengths 
which diminishes with increasing pressures and finally disappears, as the curves D' 
and E' show, when a pressure of 1000 millims. is reached. 
It is evident also from fig. 8 that the potential-spark length curves for all pressures 
greater than 1000 millims, (which is the critical pressure for tlie pressure-potential 
curve corresponding to 5 micra) will he similar in form to D' and E'. 
It thus appears that the two sets of curves, though differing widely in form for the 
lower range of pressures, yet present a resemblance as higher pressures are selected 
which becomes more and more marked. This can lie seen very clearlv from fig. 10, 
where each of the curves D' and E' has practically the same form as the curve E 
doAvn to the 5-micra line and shows no indication of not following a course similar 
to E for spark lengths below 5 micra. 
The explanation of the vertical portion of Earhart’s curves seems evident. The 
results are In reality precisely what one should expect to obtain for low pressures 
when electrodes other than ])arallel plates were used. Take, for example, a pressure of 
500 millims,, fig, 8, which Is the critical pressure for a spark length of 10 micra. 
With parallel jfiates as electrodes, it is clear that the spark potential-spark length 
curve wmdd consist of a straight line down to a spark length of 10 micra, at which 
distance the spark potential is 850 volts, the minimum spark })otential for a gas under 
normal conditions. If the distance between the electrodes is still furtlier reduced, 
the resistance offered by the gas increases and a potential difference higher than 
350 volts will be necessary In order to obtain discharge. At a pressure of 500 millims., 
therefore, a spark length of 10 micra is the one which offers least resistance. With 
spherical electrodes, for all spark lengths above 10 micra, the shortest distance 
between the electrodes is that of least resistance, and the discharge will take place 
along this line. But wheu the shortest distance between the spherical surfaces is less 
than 10, but greater than 5 micra, this distance is no longer the one which offei's least 
resistance to the passage of the discharge, and under these circumstances a longer, but 
less difficult path will lie followed. The path which offers least resistance is clearly 
the one wlilch corresponds to the minimum spark potential. It follows, then, that 
while tlie shortest distance between the electrodes is decreased from 10 to 5 micra 
and the gas is kept at a pressure of 500 millims. discharge will always occur with a 
constant spark potential of .350 volts and will follow the jiath which corresj^onds to 
this difference of potential. As Earhart’s experiments were performed with 
electrodes one of whicli was spherical and the other plane, the explanation will, in all 
probability, account for tlie ranges of constant spark potentials, which his results for 
different pressures indicate. 
The explanation which has just been given evidently requires that the constant 
spark potential corresponding to the vertical portions of Earhart’s curves should be 
3 T 2 
