426 
MR. W. R. CARR OX THE LAWS GOYERXTXG 
It is clear that all these cur^■es will lie beloAv the 5-micra curve for every pressure, 
and since a law ol proportionality applies between spark length and spark jjotential, 
at all ju’essures, it is easy to show that they fall off regularly down to the zero 
jiotential line. Ihe dotted lines a, h, c, </, shown in fig. 8, indicate the relative 
jiositions of these curves. 
In order to make a direct comparison between Earhart’s curves and those Avhich 
Ave have deduced, in fig. 9, by Paschen’s laAv, a series of each is reproduced in fig. 10. 
In this figure, A, B, C, D, and E are draAvn from the numbers giA'en in Earh art’s 
paper and correspond to pressures of 15 centims., 40 centims., 1, 2, and .3 atmospheres, 
respectively, Avhile the dotted curA-es A', B', C', D' and E’ are deduced from fio-. 8 for 
the pressures 15 centims., 40 centims., 1 atmosphere, 1000 millims. and 1200 millims.. 
resiiectiA^ely. 
For tlie higher range of spark lengths it aaoII be seen that Earh art’s Avdues are 
jiiAairiably larger than those deduced for the same pressures in this iiwestigation. 
I'his difference is especially noticeable in connection with the curves B and B', Avhich 
correspond to a pressure of 40 centims. With a spark length of 50 micra, for example, 
Earhart’s spark potential for this pressure is 625 volts, Avhile that indicated by the 
curve B' is but 470 A^olts, a difterence of about 25 per cent. This difference, hoAvcA'er, 
is exceptionally great, and extends OA'er a A'ery limited range of spark lengths. For 
distances greatei' than 1 00 micra, tlie Aailues of the spark potentials do not appear to 
differ by more than 8 oi' 10 per cent. It is evident, too, from Earhart’s diagram, 
that an irregularity exists in regard to his curve for this pressure, as it does not take 
up the position one should expect from his cui’A'es for higher and loAA^er pressures. 
A comparison of the curves corresponding to pressures of 15 centims. and 
1 atmosphere also shoAvs that the aA^ei-age difference betAveen the spark potentials for 
each of these curves, oA^er the higher range of spark lengths, does not exceed 8 per cent. 
This constant difference in the tAvo sets of results in all probability is due, at least in 
part, to the difference in the form of the electrodes used in the tAvo inA'estigations, as 
both Baille^^ and PaschexI give results Avhich shoAv that, for spark lengths of this 
order, the spark potentials obtained AA'itli spherical electrodes are in eA'ery case 
considerably liigher than those olitained Avhen the electrodes are parallel plates. 
When spark lengths slightly greater than 5 micra are reached, Earhart’s curves 
A, B, C, and D become more nearly A'ertical, and indicate that OA'er a considerable 
range of spark lengths the spark potentials remain approximately constant. It Avill 
be seen, too, that the A^ertical portion of the curves becomes shorter and shorter Avith 
increasing pressures, until finally, at 3 atmospheres, curA'e E, it disappears altogether. 
The deduced curA'es A', B, O' also present some characteristic features oA^er the 
same range of spark lengths. They each exhibit a minimum spark potential Avhich is 
reached in each case at approximately the spark lengtli Avhere the constancy of spark 
* Baiij.E, ‘ Annales de Cliimic,’ (5), vol. 25, p. 531, 1SS2. 
t Raschkn, ‘Wied. Ann.,’ vol. 37, p. 79, 1889. 
