ON THE SPARKING POTENTIALS
OF ELECTRIC DISCHARGE TUBES

r-.

V r

Tc.\'

■s

■ • - J

^ : ■]

U\' :

V ■

. gt; - ■ \'

i •

The majority of the work described in the following pages
was carried out at the Physical Institute of the University of
Utrecht during the years 1926 and \'27, under the supervision
of Prof, L. S, Ornstein, It is a great pleasure to acknowledge
my indebtedness to Prof. Ornstein for all the facilities which
he so readily placed at my disposal, for the constant interest
and help which he accorded throughout the progress of the
work, and for the great inspiration which he so unfailingly
supplies to all those who work with him.

Further, I record with pleasure my gratitude to Prof. H. A,.
Kramers, Dr, P. H. van Cittert, Dr. H. C. Burger, Dr. M. J, G.
Minnaert, and the other members of the staff of the Physical
Institute of Utrecht for their frequent help and discussion of
matters relative to the work. The writer is also especially
grateful to Mr. G. J. D. J. Willemse for his continued help
in the technical and apparatus problems and difficulties, to
Mr, H. Kuipers, who constructed all the glass work used in
the experiments, and to Mr, W. Clarkson for his frequent
discussion of the problems and his help in the preparation
of the MSS.

I have also great pleasure in acknowledging my indeb-
tedness to the International Education Board for the Fellow-
ship which enabled the work to be carried out, and to the
Armstrong College Research Fund Committee for the grant
towards the purchase of some of the electrical apparatus.

.11. V

1. ;qpnbsp;b\'vin.\'l\'VïJnbsp;lU

-vj! ,-?nbsp;.H y:M

U \'..

\'1 v\'w\' .

quot;M

■fM

■

1
\\

V-;

- i- ■

1\'

iv.

W\'

\'.a \'•

1-. i*./-« fil \'- f . f \' \'a

Electrical discharge tubes present many and interesting
problems, each requiring specialised experimental methods
and technique. Among these problems is that of the sparking
potential and its variations under different circumstances.

It is the object of the present work to examine the spar-
king potential from several aspects and to investigate its
variations under different static and dynamic conditions, to
consider it in relation to time lag effects in production of the
discharge, to external ionising factors, and to gas and elec-
trode-surface phenomena. Finally some sparking potential
theories will be considered and a photoelectric theory of the
process capable of explaining many of the exhibited pheno-
mena, will be described.

Discharge tube work suffers from grave disabilities. These
are due in the main to the emission of impurities from the
electrodes and containing vessel into the gas, the progres-
sive changes of the electrodes with time and treatment, and
to the electrostatic effects of space charge accumulated on
the walls and insulating portions of the tube.

It is impossible to overestimate the necessity for a refined
and meticulous technique in any work purporting to be a
precise contribution to the subject of electrical discharges.
Many of the phenomena are extremely susceptible to the
smallest variations of conditions of purity of filling gas or
electrodes, and to obtain exact and repeatable results is a
matter of considerable difficulty. It is therefore desirable

to outline the main technical requirements which are con-
sidered necessary, and which have been adopted in the work
of the present writer.

To avoid contamination of the filling gas by gas given off
from the walls, tubes are best constructed according to an
quot;all glass and metal-seal-inquot; technique. They can then be
quot;baked outquot; for many hours at high vacuum to remove the
condensed gases. In recent work the writer has utilised a
quot;pyrax glass and tungsten seal-inquot; technique, so that
quot;baking outquot; may be carried out at higher temperatures.

For exhausting, any of the recent techniques for high
vacuum work can be used 1) {see ref. at end of chapt,). A
twostage mercury diffusion pump backed by some form of
good oil pump is very quick and yields satisfactory results.

The progressive change of the electrodes with time and
treatment will be considered at greater length later. It is
sufficient to point out here that in order to avoid such effects
as chemical interaction under the influence of the discharge,
between the gas and the electrodes, the rare gases may
profitably be utilised. It was partly for this reason that
these gases were used for the majority of the investigations
described in this work. They were purified by Soddy\'s
method (heated calcium in quartz tube furnace) and by
means of a liquid air trap. In the later investigations char-
coal in liquid air was utilised for clearing up the residual
active gas impurities (silica gel may also be employed in a
similar manner) from helium gas.

Errors due to accumulation of space charge on the walls
etc, may be avoided by having a disposition such that the
electrodes are removed as far as possible from the sides of
the containing vessel, and separated from one another by
only a few cms.. Gas at a pressure of a few mms, is most
convenient for the experimental requirements.

There next arises the question of the electrodes. In cer-
tain cases it is necessary to investigate the electrode sur-
face changes, as progressive quot;outgassingquot; occurs. Almost

any type of electrodes are suitable for such a purpose, the
preliminary quot;baking outquot; of the tube removes much of the
condensed layers of gas but sufficient remains for many of
the progressive changes with quot;outgassingquot; to be examined.

The problem of producing quot;purequot; electrodes is one of con-
siderable difficulty and importance. To quot;overrunquot; the elec-
trodes at redheat is not sufficient for such a purpose, since
quot;sputteredquot; electrodes always possess a gas content.

In order to obtain pure metal electrodes — or rather elec-
trodes of which the surfaces were as pure as possible —
the present writer has used several methods 2).

Many years ago Warburg 3) put forward a method for the
introduction of pure sodium into discharge tubes, electro-
lytically through the glass walls. Several applications of the
method were made 4). Recently a similar process has been
utilized by Burt 5) for the introduction of sodium into metal
filament lamps by utilizing a discharge between the heated
filament and the glass walls which were partially immersed
in a bath of molten sodium nitrate at about 350° C, A suit-
ably high potential (about 300 volts) must be maintained
between the heated filament (cathode) and the glass walls
(anode).

In such a way pure sodium may be introduced upon the
electrodes of such discharge tubes as neon lamps. The neon-
discharge takes the place of the heated filament. The action
is to a large extent reversible: by making the glass walls
cathode instead of anode, sodium may bc actually removed
from the discharge tubes.

In order to investigate the properties of electric dischar-
ges in rare gases betv^een pure metal electrodes, the writer
has used a discharge tube consisting of a tungsten wire
electrode surrounded by a coaxial tungsten wire spiral, the
geometrical disposition being such that discharge always
took place from the inner tungsten wire to the surrounding
spiral. The electrodes could be heated to white heat in high
vacuum to obtain pure metal surfaces, and carefully puri-
fied argon or other gas could then be introduced. Alterna-
tively, by heating one filament in the argon a deposit of
tungsten could be condensed upon the other electrode and
its properties investigated.

Further, pure metal electrodes of various metals may be
constructed by a modification of the method of metallic
evaporation described recently in connexion with celluloid
films 6).

At the outset we shall define as quot;discharge-tubequot; any
system of two electrodes placed in a gas and between which
a spark or glow discharge may pass when conditions are
suitable. Agents which can produce ionisation of the gas
between the electrodes, or electronic emission from the
eledtrodes, arc termed quot;ionising factors,quot; and the ionising
factors of a tube at any instant are the sum total of all such
agents acting upon the tube at that moment.

There are numerous definitions of the so-called sparking
potential, and it is not easy to select one free from inherent
weaknesses. Nevertheless, if the term quot;sparkingquot; is em-
ployed it is wisest to define the sparking potential with
reference to the actual spark [or, for lower pressures, the
glow discharge, as this takes the place of the spark when

the pressure is sufficiently lowered], rather than with refe-
rence to the very small currents obtainable when the energy
suppliable by the circuit is sufficiently limited.

Adopting this point of view it may be said that, under
given conditions of pressure, temperature, and distance
between the electrodes etc., a discharge-tube is characte-
rised by a definite sparking potential if this is described in
the following manner: — The discharge-tube is carefully
protected from ionising factors — that is to say they are
reduced to a minimum — and a gradually increasing poten-
tial from a steady source is placed directly across the elec-
trodes, The potential is raised very slowly — in the ideal
case infinitely slowly — and at a definite value of this
potential, Y — vc. a spark or glow discharge occurs. This
value, vc, is termed the normal static sparking potential or
static upper critical voltage, using a more plastic nomencla-
ture 8), This is the least potential that will initiate a
discharge when there is no initial ionisation or electrons in
the tube other than the occasional ion or electron produced
in the gas or at the electrodes by small unavoidable ionising
factors such as radiations from radioactive substances, etc.
The above definition is not entirely free from ambiguity, in
that the residual ionising factors are uncertain of character
and intensity; it is more definite in nature, however, than
one which excludes all mention of ionising factors, for these
latter, as will be seen in a later part, exert a considerable
influence on the sparking potential of certain types of
discharge tubes.

If the voltage across a discharge tube is altering in some
manner not infinitely slowly with time (or in practice not
very slowly with time) the voltage at which discharge is
initiated will not in general be precisely of the value of the
static sparking potential. Indeed in some cases where rapid

change of the potential across the tube is occurring, very
serious divergences may be observed. Such sparking poten-
tials measiu-ed under changing or dynamic conditions are
termed dynamic sparking potentials or dynamic upper cri-
tical voltages, according to the notation adopted.

The divergences observed between the static and dynamic
sparking potentials in a single quot;flashquot; or discharge ¹) arise
mainly from quot;time lagsquot; in the production of the discharge,
but those observed in intermittent or oscillatory discharges
arise additionaly from residual ionisation and space charge
between the electrodes, and probably, in some cases, from
a persistance of the metastable atomic states of the discharge
gas. These phenomena will be considered later.

It has been known for a very long time that there is
usually a time-lag in the production of the discharge after
the voltage across the electrodes has attained the value vc.
This time-lag is a subject of considerable controversy, some
experimenters regarding it as of variable magnitude and
others as of approximately constant value 9), According to
Ziiber {/oc. cit.) the time-lag is determined by chance, and
obeys a probability law. Zeleny 10), for the case of dischar-
ges from points, attributes the lags to surface layers of gas
on the electrodes. Because of these lag effects and the prac-
tical methods usually adopted for the measurement of the
sparking potentials, the obtained value do not, as a rule,
correspond exactly to the static value.

It is shown later that in many discharge tubes there is an
increase of sparking potential with quot;flashingquot; or discharge.

1nbsp; quot;Flashingquot;, This is the term used for a succession of quot;flashesquot;, that
is to say discharges (usually equally distributed in .time) through a
discharge tube.

The term quot;polarisationquot; is used throughout this work to
denote this effect which gives rise to increases in the value
of the sparking potential on the passage of a discharge.

1)nbsp;Dushman, quot;Production amp; Measurement of High Vacuumquot;; Dunoyer,
quot;La Technique du Videquot;,

7} For these considerations see Taylor, Phil. Mag. 3, (1927), 368, from
which the account is derived.

8)nbsp;For these and many of the following considerations cp, Pcdcrsen,
Ann. d. Phys. Ixxi, 317, (1923). Züber, ibid. Ixxvi. p. 231 (1925);
Campbell, Phil. Mag. xxxviii. p. 214 (1919)\' Mauz amp; Secliger, Phys.
Zeits. xxvi. p. 47 (1925), For nomenclature of upper critical voltages etc.,
see Taylor amp; Sayce, Phil. Mag. I. p. 918 (1925).

9)nbsp;See ref. 8. Also Peak, quot;Dielectric Phenomena,quot; New York, 1925;
Hiller amp; Regener, Zts. f. Phys. xxiii, p. 129 (1924); Compton amp; Foulke,
Gen. Elec. Rev. xxvi. p. 755 (1923); Zeleny, Phys. Rev. xvi, p. 102 (1920).

Let us suppose that we maintEiin a voltage V across the elec-
trodes of a discharge tube of which the cathode is irradiated
by ultraviolet light or some ionising factor, in such a way that
n(j electrons are given off per second by the photoelectric
action of the radiation.

The no electrons are increased by ionisation by collision
when ihey move under the electric field in the gas between
the electrodes and a current flows whose magnitude depends
upon the voltage V, the pressure of the gas, and the distance
between the electrodes etc..

Provided V is below a definite value the discharge is unself-
sustained and dies out immediately the ionising factors are
suppressed. As V is increased the current grows at a rapidly
increasing rate until at a definite value, the sparking potential
under the given conditions, the discharge becomes selfsustained,
and if there is a sufficiently small resistance in the circuit a
spark or glow discharge occurs. This glow discharge persists
even through the ionising factors are suppressed.

The following is a method of regarding the initiation of a
spark or glow discharge 1). When the potential across the tube
is of the value of the static sparking potential or more, there
is a probability that a chance electron will, when acceleratcd
in the electric field between the electrodes, build up a suffi-
ciently strong current by ionisation by collision to initiate a
spark or glow discharge 2). The condition whether a spark
will or will not take place, will be determined by the nature
and pressure of the gas, the form of the electrodes, and the
actual place at which the electron is bom 3), Initially a

quot;dark,quot; very small current will flow owing to the presence
of ionisation due to the existing ionising factors; this then
builds up into a luminous discharge, which in the case of neon
discharge-tubes -i) and argon-nitrogen tubes 5) increases in
luminosity towards the anode, and is similar to the corona
type of discharge 6), In this diffuse type of luminous discharge
the field appears to be distributed almost uniformly throughout
the gas (with parallel electrodes), not concentrated at the
cathode as in the ordinary type of glow discharge, A space
charge is set up by the discharge and at a definite current
density the normal glow discharge or spark discharge sets in
(Seeliger and Schmekel, loc. cit.) 4),

The phenomena of this discharge, which flows before the
normal glow or spark discharge, are not, as a rule, visible *■)
because they occiu- so near to the glow or spark. This is due
to the fact that the external circuit is usually able to supply
adequate quantities of energy to initiate the spark or glow
extremely rapidly, and consequently the regime of the corona
and Townsend types of current, associated as they are with
very small energy transferences, is transitory and unstable,
Nevertheles the regime, or one very closely akin to it, may
be obtained by regulating the energy suppliable from the
external circuit, by means of a high resistance or thermionic-
valve control 7), In this way, by careful adjustment of the
energy suppliable by the circuit, the quot;corona regimequot; becomes
stable and may be studied 4). Characteristics of this type have
been studied by Seeliger and Schmekel, and Penning, for neon
glow-lamps 4). It is also the starting point for the Hoist and
Oosterhuis theory of sparking potentials 4).

It will be observed that, following an analogy of Seeliger
and Schmekel, the word quot;coronaquot; is employed here. The term
is perhaps not very good, in so far as the corona which is
obtained at higher pressures is of a rather different character.

■*) They may of coursc be studied stroboscopically, as Penning and
Clarkson have done.

However, since in the following work it is employed throughout
for the liuninous discharges of small current density which
precede the normal spark or glow discharge, or are obtained
with high resistances in the tube circuit, there can be little
ambiguity.

The features of the discharge may now be dealt with in
relation to the volt-ampere characteristic of the discharge.

At the outset we may distinguish somewhat arbitrarily
between two completely developed forms of glow discharge,

(1).nbsp;In which the positive column and its concomitant pheno-
mena are present,

(2),nbsp;In which the positive column is absent, as in discharge
tubes of which the electrodes are not more than a few cms.
apart and the gas is at the pressure of the order of a mm,
or so. Under these conditions there is the usual Crookes Dark
Space, negative glow, and Faraday Dark Space. Under certain
conditions there is in addition an anode glow.

The second type of discharge is somewhat less complicated
than the first and is the type considered here in dealing with
the volt-ampere characteristics for the higher current densities.
Nevertheless no intrinsic difference exists between the types.

The general volt-ampere characteristic for such tubes is of
the form shown in fig (1). A particular case for the small
current region is shown in fig. (2).

The experimental details for the determination of such
characteristics have been described previously 8). The discharge
tube is connected to an adjustable source of potential through
a resistance (adjustable and of the order of several megohms
maximum ¹) and the current is measured by an ammeter,
microammetcr or mirror galvanometer, according to its magni-
tude, The potential across the elcctrodes is determined by

1nbsp; Alternatively the current may be controlled by a thermionic valve,
see Penning, Phys. Zeits. xxvii, (1926), 187.

g
V.

Q Sparking •Potentials
determined
VoltMC Corresponding to
quot;Coronaquot; current«.

means of a thermionic-valve electrostatic voltmeter. It is not
permissible to place a capacity such as an electrostatic volt-
meter directly across the tube, for this alters the regime in
certain cases and may give rise to flashing or discontinous
discharges 8),

We may conveniently illustrate the phenomena of the
quot;coronaquot; currents by reference to results taken on a tube
comprising nickel electrodes 5 cm, in diameter and 15 mms.
apart at the centres. These electrodes were slightly curved
(10 Cm, radius) so that discharge took place between the
central portions of the electrodes. The filling gas was neon-
helium mixture at 7,8 mm, pressure 9),

When the potential across the discharge tube electrodes
reached the static sparking potential, fc, with a high resistance
in the circuit, a faint, diffuse glow became evident near the
anode: this luminous discharge was easily observable for
current densities o fthe order 10—quot; amperes cm. 2), 10). When
the current was increased by increase of the voltage at the
ends of the resistance it was found that the potential across
the tube remained almost of the value uc until currents of
about 5 microamperes were attained.

During this increase of current the diffuse glow near the
anode increased in brightness and began to extend. With
further increase of the current the voltage across the tube
began to fall slowly in value and the maximum of brightness
of the discharge became of some 5 mm. or so in thickness and
extended nearer towards the cathode. With further increase
of the current the discharge developed a diffuse form, convex
towards the cathode and very bright relative to the rest of
the glow. Further, when the point corresponding to A (fig. 2)
is reached the potential across the tube falls abruptly in value
and the form of the discharge alters discontinuously into a
small crescent of negative glow upon the cathode, and concave
away from the cathode (point B of fig 2), The current density
was then very much greater than in the previous type of
discharge and approached very closely the normal type with

normal cathode glow etc., and indeed the normal form is
reached on further increasing the current, and the voltage
across the tube falls to the quot;extinctionquot; or lower critical
voltage value H), The region A to B is characterised by inter-
mittence of the discharge 4), The intermittency which occurs
without additional capacity in the circuit would appear to
be due to the self-capacity of the discharge-tube system. This
is usually greater than the capacity calculated from the tube
dimensions and varies with the pressure of the filling-gas.
At the higher pressvures it was extremely difficult to obtain
a non-intermittent corona type of discharge, and indeed when
the filling-gas was air (with similar form of tubes) a steady
corona was obtainable only at pressures less than about 1 mm.

These faint luminous discharges exhibited to the right of
the region D (see fig. 1) are termed here (as we mentioned
above) quot;coronaquot; currents and the corresponding part of the
characteristic is referred to as the quot;corona characteristicquot;. It
was fotmd that within the limits of experimental error, the
potential across the tube required to start these faint corona
discharges was always of the value of the static sparking
potential (unless the electrodes were polarised, see later) even
for resistances as great as 120 megohms. The phenomena
exhibited by other types of discharge tubes are essentially
the same.

We shall now depart from the particular case of the tube
described above and consider the characteristic in its fmlher
general development.

As soon as the potential across the discharge tube has fallen
to the extinction, or lower critical value, further decrease of
resistance accompanied by a corresponding increase in current
is without influence upon the voltage across the tube. This
remains at the constant value Vf,, the extinction value, over
a comparitively large range. With increase of current however
the area of the cathode covered by negative glow increases
until the full area is utilised in discharge. This is the region
D to C of the characteristic given in fig. (1). With further

reduction of the circuit resistance and increase of current, the
voltage across the electrodes begins to rise again and finally
with zero resistance in the circuit and increasing potential the
volt-amperc curve becomes almost linear in character (region
B onwards),

The further increase of current with still greater voltages
does not concern the work in this subject and consequently
need not be considered here.

(1)nbsp;When a voltage equal to the static sparking potential
is connected to the discharge tube electrodes, discharge com-
mences and a relatively large current flows almost immedia-
tely, if there is no controlling circuit resistance. The ciurent
diminishes with decrease of potential until at a definite value
vb, the extinction or lower critical voltage, the discharge stops
abruptly and the current falls to zero.

For this part of the characteristic the writer has shown
from simple theoretical considerations 12) that the following
relation holds,

where i is the current through the tube, k is the quot;conductancequot;,
and Vg, the cathode fall of potential. It has been shown experi-
mentally for many types of discharge tubes and filling gases,
that the above relation is approximately corrcct 13), Exact
verification cannot be expected since the region of the charac-
teristic considered is one of abnormal cathode fall of potential,
and this is not taken into account in the theory. The region B
onwards to the right in the characteristic of fig. (1) corresponds
to the above relation.

(2)nbsp;With resistance in the circuit (portion B to C approx.)
we have, provided the full cathodic area is still utilised in
discharge,

Solving this simultaneous equation we obtain the following
relation for the current carried by the tube,

which is of exactly similar form to the empirical expression
given by Ryde for the Osglim lamp neon discharge tube.

The volt-ampere characteristics in the region where the full
cathodic area is employed may be distorted from that given
in the above equations, for the distance between the electrodes
frequently varies from point to point and the tube may there-
fore act as a more or less composite one having several lower
critical voltages. Furthermore, the heating effects when the
current is as great as those obtained in such circumstances,
tends to make the results tmreliable.

(3)nbsp;The vertical part of the characteristic (C to D) is a
region of normal cathode fall. The maximum current that the

be just sufficient of the cathode area employed (and covered
with negative glow) to carry this maximum current. The
current is directly proportional to the area of the negative
glow which decreases as the resistance r is increased,

(4)nbsp;The corona part of the characteristic (D onwards to
right in fig, 1) has been described above for a particular case,
and its theoretical significance will be clearer after the develop-
ment of the Threshold Current Hypothesis,

The hypothesis of Threshold currents appears to have been
applied first to the problem of sparking potentials by Apple-
ton, Emcl6us, and Bamett H), for the case of the Rutherford-
^eiger counter. In the work of these observers the threshold

We have seen that there is regime corresponding to the
corona type of current, and that this regime precedes the
normal discharge and corresponds to smaller currents. There
is thus another part of the characteristic which falls below
the normal one (see fig 1), It was further seen (fig. 2) that as
the current in the corona regime was increased the potential
across the tube fell in value from , the static sparking poten-
tial, down to the value Vb, the lower critical potential, and the
characteristic was such that if at any point corresponding to
a definite value V (lt; i^c) of the voltage across the tube elec-
trodes, the energy supplied by the circuit was increased, then
the current through the tube would increase continuously.
Considering fig. 1, ABDP represents the full characteristic 14),
AB is the normal part (corresponding to no resistance in the
circuit), and PD is the corona characteristic. The point E then
corresponds to a voltage V across the discharge-tube elec-
trodes, and a current i through it. If the energy suppliablc
from the circuit is increased whilst the voltage is maintained
at the value V, then the current will increase continuously
until the point Q on the normal characteristic is reached. It is
thus evident that V is the sparking potential corresponding to
a current i through the tube, and i may be termed the threshold
current for the potential V H). The minimum sparking poten-
tial is thus the extinction or lower critical voltage and the
threshold current is then given by the ordinate of the point C,
that is the quot;extinction currentquot;.

Experiments proved to be in agreement with the above
considerations. The method was as follows (see fig. 3). Acurrcnt
of the required magnitude was maintained through the
discharge-tube D by a battery of adjustable E.M.F., B, and
a high resistance R (1.3 megohms for the results of fig. 2).
The current was measured by means of a microammetcr M.
and the voltage across the terminals was determined by the
thermionic-valve voltmeter EvL (previously referred to). At

a given instant a voltage V (from the low resistance potentio-
meter AV continuously charging the capacity C of 0,5 micro-
farad) from an independent source was connected directly
across the discharge-tube electrodes (in the same direction as
the quot;corona voltagequot;), so as to ascertain whether a spark or

flash would occur at this voltage. The voltage was adjusted
and increased until, upon shutting the contact key K, a flash
occurred. This measured voltage then corresponded to the
sparking potential for the threshold current measured by the

The graphs of fig. 2 illustrate the results obtained. It is
seen that the sparking potential determined for the different
threshold currents agreed almost cxactly with the potential

across the tube as given by the thermionic-valve voltmeter.
Further, for this tube, as shown in the enlarged scale graph
(in fig. 2), the value of the sparking potential for threshold
currents up to 4 or 5 microamperes did not differ appreciably
from the static sparking potential It is at once seen that
with a type of characteristic such as this, which is almost
tangential to the line Y — v^ (for small ciurrents), the effect
of radiatmg the cathode with light cannot alter the sparking
potential — within the limits of measurement — from the value
v^ the static sparking value, for the currents involved due to
the photoelectric effects on the cathode are very much less
than 4 or 5 microamperes.

The form of the corona characteristic will, of course, depend
upon the shape of the electrodes etc., and in the case of certain
types of discharge-tube the ionisation produced by «-rays and
other ionising factors may be of sufficient amount to produce
a threshold current large enough to lower the value of the
sparking potential. Consequently, if such a discharge-tube is
adjusted critically, it may be utilised for the counting of
«-particles and the measurement of ionising factors 15), Apple-
ton, Emeleus, and Bamett appear to consider the above as
the method of functioning of the Rutherford-Geiger a -particle
counter. There is, however, another form of explanation of the
action of this counter, based upon the idea of resistance layers
on the point electrodes 16),

Summarising the results upon the corona currents in relation
to the Threshold Current Hypothesis we arrive at the following
generalisation.

The corona characteristic is identical with the Threshold
Current characteristic, and if in a discharge tube having a
voltage V across its electrodes, a threshold current i is pro-
duced either by external or internal ionising factors, then a
self-sustained discharge will be initiated if, and only if the
voltage V across the electrodes is equal to or greater than the
voltage on the corona characteristic corresponding to the
current I

We are now in a position to consider more fully the signi-
ficance of the volt-ampere characteristics.

The trace of the line ABCDP (that is the so called charac-
teristic) corresponds to the statical boundary conditions, and
the portions enclosed by the trace correspond to dynamical
conditions, that is to say they are only realisable when the
discharge is undergoing a flash or changing current.

(1)nbsp;AB is the statical boundary condition determined
largely by the limiting current that can be taken by the tube
when there is no resistance in the circuit (limit fixed by recom-
bination, diffusion, etc.).

(3)nbsp;DP is the corona characteristic, and is the boundary
condition characterised by the fact that on any part of it the
discharge is only just self-sustained. At points above DP the
discharge is accumulative, that is, more than self-sustained,
at points below, the discharge is imself-sustained and dies
out as soon as the ionising factors which produced the original
ionisation, are removed.

We shall sec later that we may characterise this boundary
condition by a relation similar to the Townsend relation for
the self-sustained electrical discharge. This is namely,

where [0 (v, p, i.] represents the number of ions produced by
ionisation by collision from one electron originally produced
at the cathode, for a given voltage V and current i through
the tube, y is the ratio of the number of electrons produced
by positive ions at the cathodic surface to the number of posi-
tive ions arriving there (assuming no ionisation by collision
by positive ions),

0 (v, p, 0 varies with V and t becausc of variation of elec-
trical field and space charge. The region below DP is conse-
quently characterised by the relation

and the current must necessarily die out so soon as the external
ionising factors are suppressed.

In the last section it was shown that for discharge tubes
with plane parallel electrodes the threshold current charac-
teristic was almost perpendicular to the voltage axis in the
neighbourhood of the static sparking potential. Obviously for
such tubes in which the electrode area is great, a small
threshold current cannot produce any appreciable alteration
of the space charge distribution etc, which determines the
slope of the threshold current curve. With other types of elec-
trodes such as a point anode near to a hemispherical cathode,
there may be a considerable alteration of the field space
charge with small currents, due to the relatively small mobility
of the positive ions of the filling gas, in such a way as to entail
what is effectively a decrease of the anode to cathode distance,
and lowering of the potential across the tube. Indeed, with
such a type of tube we should expect a threshold-current curve
of very small slope in the vicinity of the static sparking
potential. Such a form carmot of course be investigated in the
same way as the characteristics for plane parallel electrodes
because the part of the characteristic AB (see fig, 2) is unstable,
a very small change in current bringing about a considerable
depreciation of the voltage across the electrodes.

Irradiation of the cathode of such a tube by light or other
radiations (of sufficiently great frequency) should produce a
considerable depreciation of the sparking potential and this
ought to be applicable to the measurement of radiation
intensities.

tungsten sphere of 1,5 mms. in diameter, situated at about
3 mms, from the centre of a nickel hemispherical cathode
(about 2 cms, diameter). The tube was filled with neon-helium
mixture to a pressure of about 5,5 mms. and the electrodes
were sodiumated electrolytically.

The sparking potentials were found to be 105 volts with
the tungsten sphere as cathode, and 136 volts with the nickel
hemisphere as cathode. The tube further exhibited pronounced
susceptibility to irradiation effects. In the dark the sparking
potentials with the nickel hemisphere as cathode varied from
136,5 to 137 volts, and when the cathode was irradiated by a
parallel beam of light from a ten volt half-watt lamp (light
made parallel by system of lenses) the sparking potential fell
to between 121 and 123 volts. The polarisations (see later)
were small, not more than about a volt. With the tungsten
sphere as cathode there was only a drop of 3 volts in the value
of the sparking potential when irradiated under the same
conditions.

These results are to be expected, in so far as the threshold
current set up by irradiating the electrode of larger area must
be many times greater than that set up from the small elec-
trode, for the photoelectric effect of the radiation in the first
case is very much greater than in the second case.

Such a system could be used for the measurement of the
intensities of light and other radiations and is capable of great
sensitivity.

There remained however one serious disadvantage, the
polarisations although small were still existent and limited
the performance of the instrument, since they brought about
„zero changequot; and inconstancy of sensibility. The elimination
of all polarisation effects is therefore necessary before exact
application of the method can be made.

Since this work was carried out there has appeared two
publications of Campbell in which similar methods of using
photoelectric cells arc described 17).

certain types of discharge tube, is also applicable to some of
the forms of Rutherford-Geiger coxmter, and as pointed out
above, is adopted by certain observers.

Another interesting application of the Threshold Current
Hypothesis is to intermittent discharges or flashing. Such virork
has been carried out by Clarkson 18).

Persistence of small currents during the quot;darkquot; period of
the discharge leads to lowering of the dynamical sparking
potential, initiation of the selfsustained discharge occurring
when the current through the tube reaches the threshold value
corresponding to the instantaneous voltage across the the elec-
trodes, The dynamic sparking potential decreases as the
frequency of the discharge increases.

It is possible however, that when the dark interval between
contigluous discharges becomes sufficiently small and com-
parable with the time of life of the metastable states of the
atoms of the discharge tube gas, that persistence of the
metastable states brings about a further reduction in the

The existence of excited, ionised, and metastable states
must also, of course, accoimt to a considerable extent for the
particular forms of the threshold current characteristic, for it
is scarcely probable that the electrostatic effect of the space
charge is alone the governing factor.

1)nbsp;For these many the following consideration cp. Pedersen, Ann. d.
Phys. Ixxi, 317, (1923); Züber, ibid. Ixxvi. p. 231 (1925); Campbell,
Phil. Mag. xxxviii, p. 214 (1919); Mauz amp; Seeliger, Phys. Zeits. xxvi.
p. 47 (1925),

2)\'Marx,nbsp;Hdb. der Radiologie, 1. p. 282, Leipzig, 1920; Townsend,
quot;Electricity in Gases.quot;

4)nbsp;Penning, Phys. Zells, xxvii, p,\' 187 (1926); Secliger amp; Schmekel,
Phys. Zeits. xxvi, p. 471 (1925); cp. Hoist amp; Oosterhuis, Phil. Mag. vivi.
p. 1117 (1923).

5)nbsp;Work of W. Clarkson, not yet published. The present writer also
obtains similar results with parallel plate electrode discharge tubes with

air at pressures of less than about 1 mm. For higher pressures and point
discharges see Zeleny, Phys. Rev. xxiv, p. 268 (1925),

8)nbsp;Taylor amp; Sayce, Phil. Mag. loc. cit., Journ. Scient. Instrs. ii, p, 9
(1925); see also Penning, Phys. Zeits. loc, cit, Taylor, Phil. Mag. iii,
p. 368 (1926),

10)nbsp;Cp. Hoist amp; Oosterhuis, Phi!. Mag. xlvi p. 147 (1923). Zeleny
[Phys. Rev. xxiv. p. 258, 1924), in experiments on the discharge from
points in air (of the order of atmospheric pressure), obtained luminosity
that could bc seen by the eye for currents as low as 2 X IOt-ö amp,
(electrode a steel needle point (point negative)),

11)nbsp;This is merely descriptive. The lower critical voltage is itself sub-
ject to very considerable variations, but for the present purpose these do
not concern us. For these variations see: — Penning, Phys. Zeits. loc.
cit.; Taylor amp; Clarkson, Phil. Mag. xlix. p. 336 (1925); Taylor amp; Stephen-
son, ibid. xUx. p. 1081 (1925); Taylor amp; Sayce, ibid. I p. 916. (1925).

13)nbsp;Pearson amp; Anson, Proc. Phys. Soc. Lond. xxxiv, p. 204 (1922);
Shaxby amp; Evans, ibid, xxxvi. p. 4 (1924); Taylor amp; Clarkson, ibid, xxxvii.
p. 3 (1925).

14)nbsp;Appleton, Emeléus, amp; Barnett, Proc. Phil. Soc. Camb. xxii. 3.
p 447 (1925).

15)nbsp;Rutherford amp; Geiger, Proc. Roy. Soc. A. Ixxxi. p. 141 (1908);
Baeycr amp; Kützner, Zeits. /. Phys. xxi. 1. p. 46 (1924); Jonssen, Zeits. f.
Phys. xxxvi. 6. p. 426 (1926) (contains many references).

16)nbsp;Zeleny, Phys. Rev. xix. p. 566 (1922); xvi. p. 102 (1920); xxiv.
p. 255 (1925).

In most observations upon the sparking potentials attention
has been directed rather to the variation with different gases
and pressures, than to the variation with the nature and con-
dition of the electrodes. This is to be understood, for, according
to the classical theory, the composition of the electrodes
should be without material influence upon the value of the
sparking potential.

Recently Hoist and Oosterhuis have shown 1) by direct
experiments upon the sparking potentials in tubes having
geometrically similar electrodes of different materials, the
filling gas being neon (about 15 mms, pressure) that the
sparking potential under similar conditions depends upon the
cathode material. Thus in a special case the corresponding
sparking potentials were found to be 145 volts for magnesium,
165 for iron, 170 for carbon. For silver and copper cathodes
the values were still higher. It was determined quot;that the mini-
mum sparking potential in neon can vary as 3:1, depending
on the material of the cathode. The lowest value was found
using rubidium or caesium as cathode, the highest value was
observed with a carbon cathodequot;. This variation is as great
as the variation of the minimum sparking potentials for diffe-
rent gases.

The importance of surface films on the electrodes of discharge-
tubes has been noted by numerous experimenters. Compton
and Ross carried out some experiments on quot;charged surface
layers formed on the electrodes of vacuum tubesquot; 2), Zuber
observed a polarisation of sparking gaps 3), but attributed the
preponderant part of the effect to polarisation of the insulating
material of the gap. In the usually accepted explanation of

the Rutherford-Geiger «-particle counter, it is assumed that
a film of high resistance exists on the surface of the counter
point 4), Zeleny 5) considers the action as being due to a gas
film on the pointed electrode, and on this basis explains many
of his observations on point-to-plate discharges, lie is further
of opinion that the time-lag in the production of the discharge
is accounted for by the presence of such films 6).

It has been observed that the first spark through a tube which
has been unused for some time frequently passes more easily
and without appreciable time lag at a lower potential than
succeeding ones. On the other hand, it appears in certain cases
that the first spark passes less readily than the following 7).
These and other observations point to a change in the elec-
trodes on the passage of a discharge 7).

For the type of tube chiefly considered in this paper many
electrode fatigue effects have been signalled 8), It is known
that the sparking potential is altered and is made much more
constant when the electrodes have been purged and heated
to redness for some time by means of an electric discharge.
This stabilisation appears to result from the destruction or
modification of certain gas-layers and impurities 9), Certain
results of Ryde 10) show that surface films may play a very
considerable role in causing changes of the sparking potentials,
Clarkson in some work upon argon-nitrogen tubes H) con-
cludes that many of the anomalies inherent in such tubes
may be attributed to films of gas absorbed on the cathode.

Apart from changes caused by intense discharges which may
radically alter the nature or condition of such surface layers,
there are fatigue effects which occur in tubes where only small
energy transferences are taking place 12). Campbell, in experi-
ments on point and sphere sparking gaps in air, concluded
that quot;some very easily variable surface condition of the plugquot;
played a considerable role 13).

The above results point to a very considerable influence of
the electrode condition upon the processes of the discharge
and upon the sparking potential itself.

We shall now describe the residts of experiments upon the
change of the sparking potential with sparking or flashing. These
experiments were carried out for the most part, with tubes
prepared according to the conditions given in Chapter 1, In
some cases however, impurities were purposely introduced.

The sparking potentials were determined under definite
conditions, and it was determined that the effects were not
due to electrostatic effects on the walls of the discharge tube
owing to accumulation of charge thereon. Further, owing to
the actual form of the tubes the effects could not be due to
charges on the elcctrode supports etc..

In determining the sparking potentials the following method
was usually utilised,

A source of steady and constant potential of which the
value could be varied was placed in series with a resistance
of about half a megohm and the discharge tube whose constants
were being investigated. The latter was shunted by a capacity
of half a microfarad, and the potentials at which discharges
were produced were determined by means of the thermionic
valve electrostatic voltmeter referred to in chapt, 1, (This volt-
meter was used throughout so that the results obtained were
directly comparable with those obtained in the experiments
on the corona currents). In determining the sparking potential
the voltage was gradually raised imtil a flash occurred. The
time occupied in the process of raising the potential was usually
about 15 or 20 seconds, but provided this was not excessively
long consistent results were obtained.

1nbsp; The present account is taken largely from the Phil Mag. ill, p. 753
(1927),

sparking potential with flashing. This increase was termed
polarisation. Most tubes became stabily polarised (provided
the flashes were not very intense) after at most, some ten or
twenty flashes had passed.

The polarisation effects may be suitably illustrated by
some particular results taken on a plane parallel nickel elec-
trode tube, similar to the tube described in Chapt, 2, with a
filling gas of neon-helium mixture.

Table (I) illustrates the results obtained. The first and
third columns give the stable polarisations obtained with the
two electrodes, and the third and fourth the respective
number of flashes required to attain this stable polarisation.

It was found that when the tube had become polarised
it could be depolarisecl or recovered in two ways:

(1)nbsp;By passing a few flashes in the reverse direction
(cathode changed to anode).

The above results were obtained with the nickel electro-
des. Certain results were taken after sodium had been intro-
duced electrolytically into the tube, in order to investigate
the change in the polarisation effects.

tity sufficient to give some vapour. On examining the
sparking potentials and polarisations it was found that they
were precisely the same as before the introduction of sodium
vapour. When however, the electrodes were sodiumated (in
such a way that more sodium was introduced upon one
electrode B, than upon the other electrode A) it was found
that the static sparking potential had fallen to 192 volts for
A as cathode, and the polarisation produced an increase of
6 volts. With B as cathode the sparking potential was 172
volts and the polarisation about one volt. Results taken ten
days later showed but little alteration.

The first introduction of sodium into the side-tube
(although in sufficient quantity to be visible) made no obser-
vable difference in the results, for it was not pulled over to
the electrodes. It was evident therefore that it was not a
quot;clean-upquot; of gas or the presence of sodium vapour which
caused the differences. The seat of the phenomena must
therefore be sought upon the surface of the electrodes
themselves.

Experiments on many tubes showed similar results, but
with those likely to contain traces of active gases, large
polarisations were observed and many of these were irre-
versible.

(2)nbsp;Those which give rise to changes which are temporary
but not reversible, or are only partially reversible and are
of large magnitude.

(3)nbsp;Those which give rise to reversible and temporary
changes of a repeatable nature.

which are brought about by heavy discharges such as are
used in quot;overrunningquot; discharge-tubes for the stabilisation
of their constants 9) and which cause an actual alteration —
usually visible — of the surface of the electrodes.

It is not certain whether these effects are primarily due to
an actual alteration of the surface of the cathode or to a
quot;clean-upquot; of the filling gas or impurities. It was therefore
considered to be of interest to carry out some experiments
with a view to elucidating this.

In many cases there can be but little doubt that some of
the permanent changes are attributable to traces of active
gases introduced into the filling gas by quot;overrunningquot; etc.
for it is known that small quantities of active gas may pro-
duce profound alterations in the electrical properties of rare
gases,

A series of experiments were carried out in which the
sparking potentials were determined over a range of gas
pressures (neon-helium mixture).

The discharge tube was then quot;overrunquot; for about half an
hour, was exhausted and refilled with purified gas. The
sparking potentials were redetermined over the same range
of pressures and it was found than an increase of over 60
volts had taken place, though the gas was presumably of
exactly the same nature as previously. The tube recovered
somewhat with time but the voltages were still about 30 volts
higher than the first results.

It is to be concluded therefore as extremely likely that the
nature and condition of the cathode exert a very profound
influence on the magnitude of the sparking potentials.

(2) Under the heading (2) are the effects observed in tubes
which were likely to contain traces of active gases, The pro-
blem then arises as to whether the polarisation effects are
due to the presence of active gases as impurity in the rare
gases, or to some surface condition of the electrodes.

Experiments were carried out upon a tube filled with
neon-helium mixture and provided with a palladium wire,
which was fused into a side tube so that hydrogen could be
introduced by diffusion. The electrodes were first of all
sodiumated. Experiments showed that the polarisations were
negligeably small.

Hydrogen was introduced in small quantity through the
palladium tube, and the sparking potentials were again
determined. There was an increase of 100 volts in the
sparking potential but the polarisation was still negligibly
small. We may conclude therefrom that the presence of the
active gas hydrogen is not sufficient itself to produce polar-
isation effects. The same is probably true of other gases.

We may conclude consequently that the seat of the polar-
isation phenomena is the surface of the electrodes, and since
abnormally large and erratic polarisation effects did not
occur in those tubes whose electrodes had been thoroughly
quot;baked outquot; and quot;overrun,quot; or were quot;sodiumated,quot; it is very
probable that the effects are due to the presence of layers of
active gases absorbed or occluded at the electrode surfaces.
It is possible that many of the photoelectric effects signalled
in connexion with the neon glow-lamp 15) have arisen from
such causes,

(3) The effects classed under this heading are those dealt
with mainly in this work and appear to exist in tubes which
have been carefully treated to avoid gases of the active kind
in the neon-helium mixture and whose electrodes have been
quot;baked outquot; for several hours at 300° C, and quot;overrun,quot;

Summarising the results on the effect of the introduction
of sodium upon all the discharge-tubes used, we find the
following. The effect is dual. First of all there is a reduction
of the actual sparking potential value, and secondly there
is a very considerable diminution of the polarisation effects
observed.

As we have seen, though it is probable that many of the
permanent increases of the sparking potentials of discharge
tubes, or indeed of the slow increases, arc due to the intro-
duction of traces of active gases into the filling gas, the
reversible and rapidly changing effects are traceable to the
cathode sturface, which appears to be the chief seat of the
polarisation phenomena.

The following theory of the polarisation phenomena arose
out of the photoelectric theory of the sparking potentials
described in Chapt. 6, but it is not absolutely necessary for
the theory of the polarisation effects that such a theory shall
be adopted. All that is necessary is that some theory of
sparking potentials is adopted that stipulates that the self
sustained discharge is initiated by a process that includes an
action of the cathode surface in contributing electrons to the
build up.

According to the photoelectric theory of sparking potentials
the extra ionisation required to initiate the self-sustained
electric discharge is brought about by the emission of electrons
from the cathode, due to the photoelectric action of the radia-
tion accompanying the neutralisation of the positive ions at
the cathode surface. The sparking potential is consequently
a function of the photoelectric emissivity of the cathode and
will vary with changes of the latter in whatever manner they
may be brought about.

We shall adopt the hypothesis that the polarisation effects
arc occasioncd by the action of the discharge upon layers of
gas absorbed or occluded upon the discharge-tube electrodes.

The explanation of the rotational fatigue effect (see note 12)
becomes clear if we assume that there is a gas film at the
cathode surface and that the part of this film immediately
beneath the negative glow bccomes positively charged by the
electronic emission of the cathode and the bombardment by the
positive ions of the discharge. This electrically charged double-

layer cannot immediately dissipate its charge, and consequently
a polarisation layer is formed at the surface of the cathode.
The effect of this layer is twofold. It actually decreases the
field between the cathode surface and the edge of the negative
glow, and further it inhibits the emission of electrons from the
cathode. It is consequently easier for the discharge to pass
at an adjacent portion of the cathode that has not been subject
to the action of the discharge. The discharge may consequently
rotate and the period of rotation depends on the time for the
quot;recoveryquot; of the cathodic gas-layer.

It is necessary to consider such films in their relation to
the sparking potentials. The passage of a discharge will be
accompanied, in a way described above, by a charging-up
positively of the gas-layer at the cathode surface which is
giving off electrons and is being bombarded by positive ions
at the same time. Consequently a polarised layer is set up
after a spark or discharge has taken place. The charge of this
layer will disappear in a time depending upon the duration of
the discharge, or upon the frequency of discharge if flashes are
taking place, and upon the intensity of the discharge.

Now when a potential is placed across the polarised
discharge-tube, the existence of the electrical double-layer
will modify the conditions under which discharge occurs. In
the first place, such a layer will introduce an apparent
resistance in the circuit due to the increased contact potential
of the electrode to gas surface, and decrease the value of the
field between the electrodes for a given potential across the
discharge-tube terminals. Further, it will diminish the capacity
of the cathode metal for emitting electrons and considerably
alter the photoelectric properties of the cathode by introducing
a fatigue effect.

In most of the tubes the polarisation could be largely
eliminated, as we saw, by passing flashes in the reverse
direction. This effect is readily explainable. The positively
charged outer layer of the cathode film becomes bombarded
by electrons when the discharge is reversed, and this quickly

In those tubes showing small and reversible polarisation
effects the condition of the anode appeared to exert little
influence. This is illustrated in many results. A polarised
tube, on bemg reversed, never showed a sparking potential
value considerably less than the static value, and in the tube
upon which the results of Table (1) were taken polarisations
of appreciable value were only obtained when the electrode
which was relatively free from sodium was used as cathode.
When the sparking potential with B (the electrode upon which
there was most sodium) as cathode was 172 volts, the sparking
potential with A as cathode was 192 volts and rose to 198 volts
with flashing, that is to say there was a polarisation of 6 volts.
Now when the polarised tube was reversed so that B was again
cathode and the sparking potential immediately determined,
it was found still to be of the value 172 volts, so that an effect
on electrode A capable of producing a change of 6 volts when
A was cathode produced no appreciable change when A was
anode. From this it may be concluded that in these cases the
chief effects are located on the cathodic surface.

It is difficult to understand how such large polarisations as
those observed in some tubes can be produced if it is not due
to an effect which cither prevents the cathode from emitting
electrons so readily or alters the photoelectric emissivity of
the surface. The former explanation is applicable to the
reversible polarisations and the latter to the electrode surface
effects produced by intense discharges and quot;overrunningquot;.

The electrode surface phenomena in discharge-tubes are of
an intricate nature and the previous considerations can
merely be considered as illustrative and introductory.

It should be remarked tliat the magnitude of the polarisation
effects depends upon the intensity of the discharges and their

time-duration, upon the previous treatment of the electrodes,
and many other factors. Further, the effects are not observed
in all tubes, and this appears to be intimately connected v^ith
the nature of the gas-layer on the electrode surface. Indeed,
in one tube a slight depreciation of the sparking potential was
observed after flashing for some time. In this case the elec-
trodes had not been treated at all, and it was found that about
0,25 c.c, of gas at atmospheric pressure was given off on
quot;overrunningquot;. There was probably a considerable water-
vapour content.

Application of the Polarisation Phenomena.
Irradiation Effects with Plane Parallel Electrode Tubes.

Experiments were carried out upon the effect of irradiating
the cathodes of discharge tubes containing plane parallel
electrodes, or tungsten sphere electrodes of equal size, the
filling gas being a rare gas, usually neon-helium mixture. The
investigations were of two types, the first being upon the
sparking potentials and the time lag in the production of the
discharge, and the second upon the flashing phenomena
changes.

With regard to the first investigation it was found that the
first or initial value of the sparking potential after a long
period of rest of a discharge tube was in most cases indepen-
dent of whether the cathode was irradiated or not, and that
the time-lag in the production of the discharge was very
small. After the passage of a discharge however, it was found
that a lag in the production of discharge occurred and that
this lag was reduced by irradiating the cathode, by an amount
proportional to the intensity of the radiation incident upon
the cathode surface. At the same time the numerical value
of the sparking potential increased.

With regard to the phenomena of flashing it was found
convenient to use a discharge tube consisting of parallel
rectangular electrodes of iron (18,5 mms. by 18,5 mms,) and

3,5 mms. apart, filled with neon-helium mixture (about 78
percent neon, 22 helium.) at a pressure of about 10 mm.
The circuit shown in fig, 4 was utilised and it was observed

that there was a distinct effect on the frequency of the
flashes and of the average current by irradiating the cathode.
The effect was located at the cathode surface and was
largely produced by the blue end of the spectrum, when
ordinary light was used as irradiating agent. The graph of
fig. 5 illustrates some of the results obtained.

The above effects were traced to polarisation effects on
the cathode. They did not occur when large polarisations of
value much greater than the voltage corresponding to the
radiation employed, occurred. It was concluded therefore
that the effect was caused by the action of the light or other
irradiating agent in neutralising the electric double layers
set up at the surface of the cathode (i.e. polarisation layers).

The system is equally susceptible of being affected by
X-rays and r-rays, and is exactly analogous to some of the
forms of «-particle counter critically adjusted for counting
by flashes 16),

The action of the above described system depends upon
polarisation layers, and since these are of a somewhat erratic
and transitory nature the stability is very much inferior to
that in which the properties of the threshold current potential
curve is employed.

From the above described investigations it is perhaps
legitimate to conclude by comparison with the current results
on «-particle counters, that there are at least two different
actions involved in the various types of counter. The one
type of counter functions because of the geometrical form
of the threshold current voltage characteristic, which is of
very small slope in the neighbourhood of the normal static
sparking potential, and the other because of the existence
of polarisation layers on the cathode surface.

There have been very many irradiation effects observed
which are probably attributable to similar causes 15).

1)nbsp;Hoist and Oosterhuis, Ver. Kon. Ak. v. Wet. Amsterdam, xix. p. 849
(1920).

2)nbsp;Compton amp; Ross, Phys. Rev. vi, 3. p. 207 (1915); Ratner, Phil. Mag.
xliii. p. 193 (1922),

4)nbsp;For literature on the subject see Jonsson, Zeits. f. Phys. xxxvi. 6.
p. 426 (1926).

5)nbsp;Zeleny, Phys. Rev. xvi, p, 102 (1920); xix, p, 566 (1922); xxiv,
p. 255 (1924),

8)nbsp;See for example, Shaxby amp; Evans, Proc. Phys. Soc. Lond. xxxvi.
4. p. 257 (1925); and discussion following on this paper.

9)nbsp;Dubois, Compt. Rend, clxxv. p. 947 (1922); Taylor amp; Clarkson,
Phil. Mag. xlix, pp. 338—339 (1925); Taylor, Clarkson, amp; Stephenson,
Journ. Scient Instrs. ii. 5. p, 155 (1925),

(1924);nbsp;A. Lamberts, Phys. Zeits. xxvi. p. 254 (1925); Baeye and Kutzner,
Zeits. f. Phys. 21, p. 46 (1924); Taylor and Stephenson, Electrician. Jan. 7,

In the last Chapter consideration of the polarisation effects
led to the conclusion that the cathode stu-face modifications
were of two kinds, one of a permanent nature which changed
the electronic emissive power of the cathode, and was conse-
quently accompanied by a change in the sparking potentials,
the other of a temporary nature due to the changing up of
the surface layers and the consequent formation of electrical
double layers.

The work to be described in the present Chapter deals
with the first type of electrode surface phenomena in which
the changes of the sparking potential function are due, not
to a transient charging up of the surface layer, but to perman-
ent surface modifications.

For the purpose of investigating the sparking potential
function for pure metal electrodes — or rather for electrodes
of which the surfaces were as pure as possible — a discharge
tube of special form was constructed 1). It comprised an
electrode in the form of a tungsten wire, surrounded by a
coaxial tungsten wire spiral, the whole system being there-
fore of the same form as the filament-grid system of a ther-
mionic valve. The geometrical disposition of the supports,
etc,, was such that discharge always took place between
the inner wire and the surrounding spiral. It was quot;baked outquot;

«) The following Chapt, is derived largely from, Proc. Roy. Soc. A.
vol. 114. p. 73 (1927),

for many hours at high vacuum at a temperature of over
300° C, The tungsten wire electrodes were thoroughly clea-
ned and quot;degassedquot; by heating them strongly electrically
for six hours in high vacuum. When the electrodes had been
brought to this condition which at least approximated to a
clean, quot;gas freequot; state, carefully purified dried argon gas was
introduced into the tube to the required pressure.

The filling-gas was continuously purified by utilisation of a
quartz furnace containing heated calcium. In order to test
the quality of the gas, a discharge tube with plane-parallel
electrodes was included in the appartus. Any change in gas
quality was denoted by a change in the sparking potential
of this test-tube. Tests were made from time to time during
the work, but the sparking potential was found to be cons-
tant, consequently in the later experiments this test was dis-
pensed with, because of the excessive gas-capacity of the
test-tube.

The experimental arrangements for the determination of
the sparking potentials were similar to those used in previous
work 2), A source of constant and adjustable potential was
connected in series with a resistance of 450,000 ohms to
the discharge tube which was shunted by an electrostatic
voltmeter. The potential was gradually raised until discharge
occured. The discharge was of a feeble nature (usually of
the quot;coronaquot; type 2), and of a few microamps. in magnitude),
and was confined to the space between the wire and the
coaxial spiral.

Experiments were carried out on two tubes of similar form
and similar results were obtained in the two cases. Using S
as cathode, it was found that the first discharge took place
at a definite potential and the succeeding ones at progressively
decreasing potentials. The value of the sparking potential

S denotes tungsten spiral, W coaxial wire. The numbers
at the side of the sparking potential values give the number
of the discharge in the time order in which they occurred.
Tungsten electrodes quot;formedquot; in vacuum. Pressure of argon
3,94 mm. Diameter of tungsten wire 0,3 mm,; diameter of
spiral, 6 mm,; length, 16 mm,; number of turns, 10,

poten- ; Wnbsp;cathode. (16) 245 (17) 246 (18) 247 (19) 247.5
tials. I Remained fairly steady at this value,

decreased gradually vmder the action of the discharge to a
minimum value after which it increased again. Table H is
illustrative of the results obtained with one lube. In the results
given the first discharge took place with S as cathode, so that
the value (2) given for W as cathode does not represent the
actual value for a vacuum-formed surface on W, In the
example given the values for the sparking potentials with both
S and W as cathodes finally attained values approximating
to the initial ones. This, however, was by no means always
the case. There appeared to be great differences in behaviour
according to conditions, and the previous treatment of the
wires 3). In all cases, however, it was found that the sparking
potentials for both S and W as cathode were higher initially
with the quot;clean timgstenquot; electrodes (the initial sparking
potential was constant and definite) and decreased with
discharge to a minimum value which varied according to
conditions.

After this point, in almost all, though not quite all, cases,
a considerable but slow increase in the sparking potential
values occurred. In some few cases the final values were even
higher than those characteristic of the quot;clean tungstenquot;.

A series of experiments were undertaken in which the
tungsten electrodes previously quot;formedquot; in high vacuum were
reheated for various periods of time in argon gas which was
continuously purified.

The results obtained were almost the same as those for the
vacuum-formed electrodes, except that the initial sparking
potentials were usually lower than for those. Nevertheless, in
many instances, the values approached those for the vacuum-
formed surfaces. Tables III and IV give some of the results
obtained.

Treatment of Tube.

264nbsp;S heated dull red for 0.5 minutes in the gas.
261 S heated strongly a few minutes in the gas.
264 S and W heated strongly a few minutes in

W heated strongly a few minutes.
S and W heated a few minutes together,
W heated very strongly for a few minutes,
W heated a few minutes.

The result of heating appears to be as follows: — If S was
heated, then its final condition was very similar to that which
it wouid have attained in vacuum, but the condition of W was
unaffected except for an indirect effect due to the heating it
underwent by conduction and radiation. If, however, S was
heated very strongly, tungsten from its surface evaporated
on to W and formed a new surface there. Whatever the nature
of the surface layers that were formed under the action of
the discharge they were modified or done away with when
the wires were heated in the pure argon gas. The sparking
potentials were not quite the same as those attained by heating
in vacuum, so that it must be assumed that the heating did
not drive off so much of the surface layers in the gas as in
vacuum, or alternatively, that the change in the surface
structure was not so great as that obtained by heating m
vacuum.

Another important point which will be considered later,
is that the actual relative values of the sparking potential for
S and W as cathodes depended largely upon the condition of
the electrodes and not merely upon tiie geometrical distribution
of the electiic field between the wire and the coaxial spiral,
a result which follows tiie assumption, until recenUy held, that

Volts,
249
241
239
251
245
255
251

the sparking potential was independent of the nature of the
cathode surface 4). By comparing the results in Table IV for
W as cathode, with those in Table III for S as cathode, it is
seen that normally, when the electrodes were most probably
in the same condition, the sparking potential was lower for W
as cathode than for S as cathode. This is in accordance with
the usually accepted results 4), In cases where S and W had
undergone different treatment, and were, therefore, presum-
ably of different surface conditions, the sparking potential for
W as cathode could be less than, equal to, or greater than,
that for S as cathode, according to circumstances. Table V
gives some illustrative results.

The condition of the anode was found to exert little, if any,
action on the values of the sparking potential. We may
conclude, therefore, that the value of the sparking potential
depends upon the surface condition of the cathode.

Sodium was introduced electrolytically through a side tube
provided for the purposes), and was evaporated on to the

tungsten electrodes. After the tube had been rested for a
few hours, the sparking potentials were redetermined. Pre-
vious to introduction of the sodium the values were 265 volts
with S as cathode, and 251 volts with W as cathode. These
values were those determined after S and W had been
heated to yellow white heat for ten minutes and then rested,
When the electrodes were sodiumated the values were 122
volts with S cathode and 106 with W cathode. The falls of
the sparking potential values were consequently great. The
electrodes were then treated in a manner described in
Table VI, in order to drive off some sodium from one or the
other electrode. After each heating the tube was rested
several minutes and the sparking potentials were then
redetermined,

From the table (second line) it is seen that when W was
heated to a dull-red for 1,5 minutes the sparking potential
had risen by 108 volts (W cathode). The value attained was
not that characteristic of the original tungsten electrode,

so that it is evident that the heating had not entirely repro-
duced a tungsten surface. At the same time S had evidently
become heated to a slight extent by conduction and radiation,
and some of the sodium had evaporated. This brought about
an increase of 14 volts in the sparking potential with S as
cathode. In the next operation S was heated for an extre-
mely short time. It was found that the sparking potential (S
cathode) had risen a further 6 volts, whilst the value for W
had fallen from 214 volts to the value 98, showing that sodium
had evaporated from S on to W, The table shows the contin-
uation of similar treatments. It is seen that when S was
heated for 10 sees, or more, that W also became heated by
conduction, etc., and lost some of its sodium, so that the
sparking potential increased in value.

The effect of introducing sodium upon the cathode is to
diminish progressively the sparking potential as the amount
of sodium is increased, until a constant value is attained
when the electrode is completely sodiumated. The heating
reduces the quantity of sodium per unit area. We may as-
sume that a film of a few atoms in thickness will act in all
ways as a sodium electrode. With an insufficient quantity
to produce such a film there will be quot;bare patchesquot; and the
electrodes will be quot;mixedquot; ones.

The above described results are in accordance with those
given in the previous Chapter,

Summarising we may conclude that: —
The Sparking Potential is a function of the nature and
condition of the cathode surface, and varies continuously with
changes of the latter, depending upon the mean composition
of the quot;workingquot; part. In other ivords the Sparking Potential
is a function depending upon a gross effect of ions acting upon
the cathode surface.

We may reasonably attribute the effects of the change of
the sparking potential in the case of the tungsten electrodes

to changes in the surface layers of gas on the cathode. This
was the position maintained in previous work on the electrode
surface effects. It is, of course, also conceivable that many of
the above described effects were due to changes in the surface
structure of the metal of the cathode; in any case, such changes
would be accompanied by a corresponding change in the photo-
electric emissivity and it is to this change that the alterations
in the value of the sparking potential are attributed.

The results, indeed, axe very analogous to some of those
obtained for the variation of the photoelectric emissivity of
metals 3),

3)nbsp;Compare with experiments on the photoelectric emission of metals
as influenced by temperature, e.g., Welo, Phil. Mag. vol, 2, p, 463 (1926),

5)nbsp;For method see Taylor, loc. cit. Cf. Warburg, Ann. d. Physik.
vol, 40. p, 1 (1890); Burt, Phil. Mag. vol, 49, p, 1168 (1925)| Journ. Opt.
Soc. vol, 11, p, 87 (1925), See also Taylor, Journ. Scient. Inst. vol. 4,
p. 78 (1927),

The actual mechanism of the initiation of the self-sustained
electric discharge is still obscure.

With regard to the development of currents in the regime
before a self-sustained discharge is obtained, experimental
evidence is fairly decisive quot;in showing that ionisation is
principally due to the direct effect of collisions of electrons
with molecules (or atoms) of the gasquot;. Quite how this ionisa-
tion by collision is brought about is a matter of speculation.

According to the Classical Theory of Towsend the ionisa-
tion effect of an electron is characterised by a coefficient «
which is defined as the number of electrons produced per
cm,, under a definite electric field and gas pressure etc,, by
collision 1),

For parallel plate electrodes in which the cathode was
irradiated by ultraviolet light or by X-rays, Townsend has
shown that for a considerable range of distances x, between
the plates, when the electric force X is constant,

where n is the number of ions arriving at the cathode and
n,) is the number of electrons produced originally by the ac-
tion of the ionising factors upon the cathode. This relation
arises naturally out of the Townsend Theory of ionisation by
collision, it further gives us one requirement for a satisfac-
tory sparking potential theory, namely such a theory must

account for the fact that the quot;currents increase in geometri-
cal progression as the distance increases in arithmetic pro-
gressionquot;.

On the other hand we may consider the phenomena from
the point of view of the results obtained in recent years in
the region of quantum physics by Franck and Hertz and
other workers. According to these ideas the collision of an
electron possesses an energy equivalent to the resonance
potential or ionising potential of the gas considered.

It is not the purpose of the present work to consider such
processes in any detail. In the case of diatomic, electroposi-
tive and electronegative gases, the complexity becomes such
as to preclude any satisfactory treatment at all, for almost
nothing is known about the mechanism and process of collis-
ions in such cases. With the noble gases conditions are much
simpler; they are monatomic and without large residual
Itomic fields of their own. It is therefore preferable for the
simplification of conditions to utilise such gases in most work
upon sparking potentials.

Electronic collisions against atoms of the rare gases are
almost entirely elastic until some critical velocity equivalent
to a resonance or ionising potential is attained. Consequently
an electron in the electric field between two electrodes may
move in any way from atom to atom and will acquire an
energy equivalent to the fall of potential in the direction of
the field that it has experienced, provided this fall of potent-
ial does not exceed a certain critical value. When the elec-
tron has in this manner attained an energy eV,^ where e, is
the electronic charge and V,^ the ionising potential of the
gas, it will ionise an atom on collision and give rise to a new
electron. These new electrons will in turn produce more
electrons under similar conditions. We may assume conse-
quently that in the space between two charged electrodes
an electron will ionise at distances characterised by a drop
of voltage equal to the ionising potential.

for many questions arise, which up to the present remain
unanswered. For instance we may ask what is the propor-
tion between collisions occurring when the electron has an
energy equivalent to the resonance potential and those oc-
curring when the energy is equal to the ionising potential.
Also what part does the radiation emitted by such processes
play in the mechanism?

Such a process as envisualised above is embodied by
Hoist and Oosterhuis in their theory of the sparking potenti-
als 2). They assume for an hypothetically ideal gas:

(1)nbsp;quot;An electron loses no energy whatever in collisions as
long as its velocity is below that corresponding to the
ionising potentialquot;,

(2)nbsp;quot;An electron will ionise as soon as its velocity is equal
to the ionisation potential Viquot;.

which takes the place of equation (6) of Townsend\'s Theory,
g is the number of ionising potentials in the total voltage
across the discharge tube electrodes. This equation accounts
for the geometrical progression form of increase of current
with distance between the electrodes. The equation (7) was
developed for the case of an ideal gas. In practice energy is
lost by collisions. Hoist and Oosterhuis consider this loss
and introduce a second approximation to take into account
the effects. The resulting expression is complicated and does
not lend itself readily to calculations.

There is little doubt that up to the present the Townsend
formula for ionisation by collision of electrons is superior to
others from the point of view of applicability to working
conditions of gases in bulk, but whether it satisfactorily des-
cribes the processes involved is open to doubt.

Assuming that the development of currents in the regime
preceeding a self-sustained discharge is adequately repre-

sented by one or other of the above theories or their modifi-
cations, the question remainsquot; which of the other modes of
generating ions contributes the additional effect required in
order to explain the disruptive dischargequot;.

The source of this additional ionisation is generally sought
in an action of the positive ions liberated by the ionisation
of the atoms or molecules of the gas by the electronic collis-
ions, or in a photoelectric action at the cathode of radiations
produced in the bulk of the gas 3).

The question of the production of ionisation and electrons
by the action of positive ions is one which has engaged much
attention and in which very many contradictory conclusions
have been reached.

Our attention here is more especially directed to the ac-
tion of positive ions of energy corresponding to not more
than a few tens of volts, traveirsing gases, for these are the
conditions corresponding to the regime of the self-sustained
electric discharge in gases. Notwithstanding, it is of interest
to consider somewhat generally the experimental findings in
this branch of research.

The ionisation of gases and the production of intense
secondary electronic emission from solid targets, by the
bombardment of «-particles is well known.

The ÛC rays from radium C, possess energy corresponding to
about 2,000 kilovolts, and represent the linut of energy
possessed by any positive ions.

They shatter any atoms which they encounter, and in air
at atmospheric pressure form several thousands of ions in
every millimetere.

Canal rays (of maximum velocity about one hundreth of
that of the « rays from radium C) also produce ionisation by
collision against gas molecules and cause an emission of
electrons from metal targets, Villard concluded that cathode
rays vk^ere formed by bombardment of the cathode by positive
ions of the discharge 4).

Seeliger 5) in some quantitative work, has concluded that
the number of ions produced by positive rays is small compared
with the energy of the positive particles, and that the number
is almost independent of the velocity of the rays and of the
pressure of the gas.

The emission of positive ions from heated vrires and salts
has received a considerable amount of attention in reference
to the influence of the pressure and nature of the gas in which
they are enclosed etc.. Definite researches to detect ionisation
by collision of the emitted ions against the molecules of the
gas and by impact against metallic electrodes have been
carried out with varying and conflictory results,

Klein 6) detected a secondary emission from a nickel electrode
upon which positive ions of 50 volts velocity impinged. The
emission became 22 percent of the positive ion current value
when the potential was raised to 380 volts.

In some recent experiments Jackson has employed a
beam of K ions from the iron catalyst source discovered by
Kunsman 8). quot;The secondary emission of aluminium, nickel
and molybdenum under a variety of surface conditions up to
1,000 volts was measured. Heat treatment usually involved a
diminution of the secondary electronic emission. The secondary
electrons emitted were of low speed, a retarding field of a
fraction of a volt was enough to stop nearly all of themquot;.

Stark 9) in examining the current between a heated carbon
filament and metal cathode in air, observed an effect of
increase of current which he attributed to ionisation by collision
of the molecules of the gas by the positive ions from the carbon,
Pawlow 10) carried out experiments from which he concluded
that positive ions produce ionisation by collision in gases.

Franck and Bahr H) obtained ionisation in air and hydrogen
below the ionisation potentials of these gases. They also found
that the lowest voltage for which ionisation could be detected
was smaller for larger intensities of the source of positive ions.

More recently Horton and Davies 12) have conducted
experiments upon the positive ions emitted by a tantalum
filament in helium.They detected an increase of ionisation with
positive ions which had fallen through potentials as low as
20 volts and attributed their results to the electronic emission
from the walls of the ionisation chamber by bombardment of
the positive ions. Further they are of opinion that positive ions
do not produce further ionisation by collision when accelerated
through potentials of 200 volts,

Saxton 13) has carried out experiments utilising the positive
ions of hydrogen. The results were interpreted as giving
quot;definite evidence of ionisation by positive ions accelerated
through potentials as low as 18 voltsquot;.

In a recent paper quot;On the electric discharge through gases
at very low pressuresquot; Sir J, J, Thomson 14) deals somewhat
extensively with the question of ionisation by collision of
positive ions and discusses several objections to the hypothesis
that low speed positive ions are able to ionise by collision
against gas molecules. He discusses other methods by which
the ionisation in the discharge, which is often attributed to the
above action, may be produced, and puts forward an alternative
and very interesting hypothesis according to which the positive
ions liberate photoelectrons by the action of the radiation
emitted when they are neutralised at the cathode surface.
Continuing work in this direction Sir J. J. Thomson 15)
examined directly the radiation effects due to the streams
of cathode and positive rays through a gas. It was shown that
the passage of the positive gas ions through the gas itself
produced radiation from the gas. Further, it was found that
the radiation produced by the positive rays striking against
metallic targets was similar to that produced by the passage
of the positive rays through the gas. The radiation was very

soft, not exceeding a quantum of the order of the ionising
potential of the gas.

Work of a similar nature on the radiations of electric dis-
charges in the rare gases helium, neon, and argon, at voltages
across the electrodes of up to 400 and pressures of the order
of a few mms. has been described by Dauvillier 16),

The conclusions differ somewhat from those of Sir J, J,
Thomson, It is attested that no radiation attributable to the
neutralisation of the positive ions at the cathode surface was
observed. The impact of the positive ions of the discharge
against the cathode is supposed to liberate electrons by
electrostatic attraction.

Hoist and Oosterhuis n) discussing work upon discharges in
rare gases, notably neon, regard the production of ionisation
by collision of the positive ions in the gas as improbable and
attribute the quot;building upquot; of the discharge to electronic
emission from the cathode by the electrostatic attraction of
the positive ions on the electrons in the cathode, and describe
results in support of their theory. This hypothesis is also one
of the bases of their theory of sparking potentials 2),

Townsend in a long series of researches carried out over
many years maintains the position that ionisation by jollision
of positive ions against molecules or atoms of gas, even though
these ions are of slow velocity, is the normal source for that
additional ionisation required to account for the self-sustained
electrical discharge in gases 1 amp; 3),

There are three main theories to explain the ionisation
produced in gases by positive ions. The oldest is the now
classical theory of Townsend of ionisation produced by collision
of the positive ions against the molecules of the gas in which
they are moving. This ionisation is brought about in much the
same manner as that by the electrons. The second theory is
that the effect attributed to ionisation by collision of low

velocity positive ions is really due to secondary photoelectric
action of the radiation emitted by the gas molecules or atoms,
and the third is that the quot;extraquot; ionisation observed is due to
the electrons produced by the bombardment of the electrodes

(1)nbsp;In which electrons are produced by the quot;knocking outquot;
of electrons by the impact of the positive ions against the
cathode,

(2)nbsp;In which the electrons are supposed to be given off by
the photoelectric effect of the radiation emitted during the
neutralisation of the ions impinging on the cathode,

(3)nbsp;In which the electrons are produced by the electros-
tatic attraction exerted by the positive ions upon the elec-
trons in the surface of the cathode,

(1), (2), and (3) may be classed together under the heading
quot;electronic emission due to positive ion impactquot;.

The first and second of the main theories described above
have been treated somewhat extensively by Townsend 3) and
it is the object of the present work to deal more especially with
the three divisions of the third main theory.

The experimental evidence given above leaves little doubt
of the fact that the action of positive ions impinging upon
metal targets gives rise to secondary emission of electrons.
There are of course disagreements as to the magnitude of
such an emission and as to the kinetic energy requirement
for positive ions to produced such an effect, the mechanism
of the phenomena and its dependence upon the velocity of
the particles.

sion. is produced by the impact effect of the ion against the
atoms in the surface of the target. By virtue of the kinetic
energy attained during their acceleration in the electric field
between the electrodes, and that due to the electrical image
attraction of the positive ions, these ions are able to ionise
the atoms or knock out conduction electrons from the sur-
face space lattice. The magnitude of such an action will
depend upon the kinetic energy of the positive ion, and will
increase with the latter. Further, since the majority of the
electrons are likely to be liberated in the direction of the
motion of the positive rays, the secondary electronic emis-
sion would not be expected to be large for small velocity
ions, and indeed when the velocity is below a certain critical
value no electron emission will be detectable,

Jackson\'s results which were quoted above are interesting
from this point of view. quot;The secondary electron emission
could not be detected (was less than 0.5 percent) at positive
ion velocities less than 200 volts for Al, 300 volts for Ni, and
600 volts for Mo after heat treatment. The secondary emis-
sion increased from these values to 7 percent for AI, 4,7 per-
cent for Ni, and 3.8 percent for Mo, at 1000 volts. Without
heat treatment the emission was detected at lower voltages
and reached about double the above values at 1,000 voltsquot;.
From these and other results we may conclude that the
number of electrons liberated from the cathode of a discharge
tube by such a process, by the very slow ions involved in
such conditions as are usual in the initiation of a small self-
sustained electric discharge, is very small. In some cases the
small emissions mentioned above may be sufficient to explain
the extra source of ionisation, but in the majority of cases
it appears to be insufficient. Further, it should be pointed
out that the emissions observed in such experiments as those
of Jackson, may be attributed to the process described
under heading (2). Potassium ions on neutralisation emit
radiation of a maximum quantum that is much inferior to
that from such ions as the rare gas ions etc, and conse-

This is the hypothesis put forward by Sir J. J. Thomson,
and described above in the reference to this work.

We should expect that the photoelectrons produced by
the radiation accompanying the neutralisation of the posi-
tive ions at the surface of the cathode will be largely inde-
pendent of the velocity of the positive ions over a conside-
rable range. This is in agreement with the experimental findings
given previously. As the velocity of the positive ions in-
creases however, the case may be different. We may imagine
that, provided the velocity of the positive ion is sufficiently
small, the photoelectric effect is simply due to the single
neutralisation of the ion, but when the velocity is sufficiently
increased (due to increase of the surface field) we may
consider with Sir J. J. Thomson that quot;A positive ion striking
against the cathode may alternate from the charged to the
uncharged condition, if it has much energy, many times be-
fore it loses its charge for the last time; each change
from the charged to the uncharged state would be accomp-
anied by the emission of radiation.quot; Thus with the higher
velocity positive ions the effect may become considerably
enhanced.

Further, the number of electrons emitted by the photo-
electric effect will depend upon the magnitude of the quan-
tum of radiation emitted by the ion on neutralisation. The
maximum value of this quantum will, for a given ion, corres-
pond to the energy associated with the ionising potential. We
should expect consequently that the photoelectric emission
produced by the positive ions from the rare gases for which
the ionising potentials are high, will be greater than that for
the ordinary gases.

Futher, it is to be expected that the emission will increase
with the surface field.

The third theory is that the secondary electron emission is
produced by the electrostatic attraction which is exerted
by the positive ions upon the electrons in the surface layer
of the cathode. On this hypothesis the secondary emission
would be largely independent of the velocity of the positive
rays but would depend upon the intensity of the beam im-
pinging upon the target. We shall call this theory the quot;Auto-
electronic Theoryquot;,

Recent work carried out upon quot;autoelectronic emission,quot;
that is electronic emission from a metal electrode due to the
maintainence at its surface of an electrostatic potential suf-
ficiently high, are very interesting in the light which they
bring upon the possibility of the process described above.

Of these results it is sufficient to quote those of Millikan
and Eyring 18), which are relative to our purpose. Gossling
has obtained very similar results 19),

Millikan and Eyring investigated the currents quot;from thori-
ated tungsten filaments in vacuum due to radial fields up to
2 X 10quot; volts per cm,quot;. The results showed that the current
increased from 10quot;\'^ to 10~^ amps, with increase of the
field from 400 to 1100 k,v. per cm,, and clearly indicated
that the currents came only from a few active surface spots.
The magnitude of the current was found to depend upon the
previous treatment of the wire and was largely independent
of the temperature over a wide range. It was suggested quot;that
the field currents are due to conduction electrons pulled
from minute peaks on the surfacequot;, and at these places the
field must be considerably in excess of the average value
over the wire surface.

If we make a rough calculation of the order of distance
between a positive ion and electron required for the positive
ion to exert a force of a million volts per cm,, it is readily
seen that the order is that of the electronic orbits of the
atom and is such as makes it improbable that the electron

shall escape from becoming part of the atomic system of the
ion by neutralising it.

This process will emit radiation and give rise to the effect
described in Theory (2). We may conclude therefore in such
circumstances as we have in near consideration, namely
those corresponding to gases at several mms. pressure and
voltages of low value, that the electrostatic attractions exer-
ted by the positive ions which impinge against the cathode
are insufficient to cause electronic emission except by an
indirect process.

We may summarise by concluding that so far as the pheno-
mena of the self-sustained electric discharge is concerned
the source of the additional ionisation is to be found either in
the process of ionisation by collision of positive ions described
previously, or in the processes described under heading (1)
and (2) of the theory of Electronic Emission due to Positive
Ion Impact, and further, that there is a balance of probability
in favour of the theory described under heading (2), that is
the Photoelectric Theory of the the emission of electrons due
to positive ion impact,

In the last Chapter we developed different hypotheses which
can explain the ionisation produced by electrons and positive
ions. It now remains to select from among these hypotheses
those best capable of providing a Sparking Potential Theory
that will account for the various phenomena observed in the
initiation of self-sustained electrical discharges. Such a theory
must be capable of explaining Paschen\'s Law (Sparking
potential is a function of the product of the pressure of the
gas and the distance between the elcctrodes for plane parallel
electrodes), the phenomena observed for the case of plane
parallel electrodes, for discharges between a wire and coaxial
cylinder, and the various electrode surface effects associated
with discharge initiation.

The theories developed in the present chapter are more
particularly for the case of pressures above those corresponding
to the minimum sparking potentials, but their extension to
other cases is evident.

Initially in formulating a theory we must account for the
ionisation by collision produced by electrons, and this involves
the acceptance of either the classical theory of Townsend (or
one of its modifications), or a theory based on quantum
considerations such as was adopted by Hoist and Oosterhuis.

For empirical purposes Tov^rnsend\'s theory is satisfactory
and easy to handle, and in default of further knowledge of
the quantum reactions occurring in gases in bulk, it is perhaps
the most convenient to adopt.

produced by the positive ions, which we supposed were the
effective agents in this direction, *)

It was shown that there are several possible explanations
of the ionising action produced by positive ions, and, a priori,
we cannot be sure that in all cases of initiation of the discharge,
the process is identical. We also do not know definitely that
only one action of the positive ions is effective in any one
case. There may be indeed, cases where the initiation proceeds
from the joint effect of one or more of the actions described
in the previous Chapter; it is more probable however that one
effect alone will be prédominent, and consequently the foll-
owing generalisation may be made: —

The initiation of the self-sustained electrical discharge will
be brought about by that particular action of the positive ions
that is most easily accomplished under the conditions of
experiment.

The statement is somewhat imrigid in wording but its
meaning is clear. If in a given case for example, ionisation by
collision with the molecules or atoms of the gas is more readily
produced than photoelectric emission from the cathode by the
action of the positive ions, then the former process will be
chiefly responsible for the initiation of the discharge, and
vice versa.

An exact theory would therefore emerge out of a knowledge
of the relative probabilities of the processes summarised in
the previous Chapter.

In the Townsend Theory both electrons and positive ions
are assumed to produce ionisation by collision with the mole-

■*) There is also the theory mentioned in Chapt. 5, that the quot;extraquot;
ionisation is produced by a photoelectric effect of radiation emitted from
the bulk of the gas. This is one of the cases considered by Townsend
Owing to the lack of definite evidence in favour of such a theory it is
not treated here.

cules of the gas. If iIq electrons are liberated from a plane
metal cathode and move in the imiform field between parallel
plates, the nmnber of ions n, gathered to the anode is, according
to this Theory,

quot;where « is the average number of molecules ionised by an
electron in moving through one cm, of the gas in the direction
of the electric force, /3 is the average number ionised by a
positive ion, and x is the distance in cms. between the plane
parallel electrodesquot;. Obviously then the condition for a self-
sustained electric discharge 1o be produced is given by the
relation,

and the sparking potential is consequently Xx where X, is the
electric field between the plates.

The theory of Townsend accounts for Paschen\'s law and
for the phenomena observed in the case of discharges between
a wire and coaxial cylinder.

The Townsend Theory of ionisation by collision for negative
ions or electrons is not universally adopted, but leaving this
aside the objections that can be raised against the Townsend
sparking Potential Theory are: —

(1)nbsp;The hypothesis that low speed positive ions produce
new ions by collision with the molecules of the gas in which
they are moving is, as we saw from the results described in
the previous Chapter, open to very grave doubt.

(2)nbsp;According to this theory the sparking potentials should
be independent of the nature and condition of the cathode
surface, whilst in practice it is found to be largely dependent
upon these,

Dubois 2) has modified the Townsend Theory to take into
account an additional action of the positive ions in liberating

electrons from the cathode. The ratio of the number of ions
impinging upon the cathode to the number of electrons set
free by them is characterised by a coefficient r. and the revised

In the last Chapter a description of the hypothesis adopted
by Hoist and Oosterhuis for the ionisation by collision of

(2)nbsp;quot;No positive ion ionises a gas molecule by collisionquot;.
The mechanism whereby the positive ions are supposed to

liberate electrons from the cathode is that described in
Theory (3) of the Electronic Emission due to positive ion
impact (previous Chapter).

According to the theory of Hoist and Oosterhuis the number
of positive ions produced by an electron originally given off
from the cathode is (29—1), (equation (7), Chapt, 5). The
condition for the discharge to set in is consequently given by
the relation,

93 being a particular definite value of g.
The sparking potential v^ is then given by the relation,

For the case of the hypothetical gas postulated by Hoist
and Oosterhuis the sparking potential would be independent

The relation of equation (11) is arrived at on the assumption
that no energy is lost in collisions of the electrons against the
gas atoms at.velocities below that corresponding to the ionising
potential. On introducing an approximation for the loss of
energy on collision, the following relation for the sparking
potential is arrived at,

in which a, is the electrode distance in cms., p, is the pressure
of the gas in cms, hg., K is the M,F,P, of a molecule of the
gas at N.T.P..

quot;The departure from the ideal gas is due to energy losses
of the electrons, If the number of collisions is small these
losses will be less important. That is the reason why the
minimum sparking-potentials for gases are not very far apart,
the difference in sparking-potcntial increasing with pressure
and electrode distancequot; 3).

Hoist and Oosterhuis have shown that relation (12) holds
approximately for neon.

The theory has the advantage of being capable of explaining
the variations of the sparking potential with the nature of
the cathodic surface, and of many of the observed surface
phenomena. It is unfortunately of very limited application.

In a recent paper Townsend 4) criticises the theory and
brings forward certain arguments against it.

In the light of recent experimental evidence it would appear
doubtful whether the production of electrons from the cathode
can be attributed to an electrostatic attraction effect. It must
bc noted however that the form of the theory remains

unchanged if either of the alternative hypotheses (1) or (2)
as described in Chapt, 5, are adopted.

The Photoelectric Theory of sparking Potentials arises
naturally out of the hypothesis of Sir J. J. Thomson described
in Theory (2) of the hypotheses of electronic emission due to
positive ion impact, Chapter 5. It was also an expression of
the conviction that it is necessary to attribute a not imimportant
part of the mechanism of the discharge to radiations given
out by the gas.

If we assume that for n^ electrons produced originally at
the cathode, (by the ionising factors), n„ (p (V, p,) electrons
are gathered to the anode, where cp (V, p.) is a function
depending upon the voltage across the tube and the pressure
of the gas etc., then the number of positive ions arriving at
the cathode is

. (13).

n = n.

The neutralisation of these positive ions at the cathode
surface is accompanied, we shall suppose, by the emission of
radiation (see previous Chapter) some of which falls upon the
cathode surface and produces the emission of photoelectrons.
Let the ratio of the number of electron given off from the
cathode by the photoelectric effect to the number of the
positive ions neutralised there, be r- Then we have the following
condition for the production of a self-sustained electrical
dischsu-ge.

cp (V. p.) - 1

r n.

equation (15) resolves itself for the case of plane parallel
electrodes, into the following relation,

where X, is the electric field between the electrodes and x,
is the distance apart.

Alternately if we accept the Hoist and Oosterhuis theory
of the production of ionisation by electronic collision the form
of their expression for the sparking potential remains exactly
the same (equations (11) and (12),),

Equation (3) is precisely the same as the equation given by
Townsend, except that y has a different significance 4).

The variation of the function y, which is a measure of the
photoelectric emissivity of the cathodic surface for the radiation
accompanying the neutralization of the positive ions of the gas,
would entail a whole manifold of possible variations of the
sparking potential function v^. 7 would be variable according
to the condition and mean composition of the quot;workingquot; part
of the cathode surface, and would be subject to variations with
the gas-to-metal potential changes, and the transient electric
double layers set up by electrical charges on the cathode
surface.

The results, indeed, are very analogous to some of those
obtained for the variation of the photoelectric emissivity of
metals 6), It is hoped to continue experiments in which both
y and Vc will be measured simultaneously.

It will be noticed that throughout this Chapter no specific
reference has been made to any dependence of the sparking
potential upon the magnitude of the ionising factors. It was
shown in Chapter 2 however, that the sparking potential is
dependent upon the magnitude of the threshold current.

The equations given above (with the exceptions of (13), (14)
and (15).) must be taken as giving the static sparking potential
as defined in Chapter 1.

that the Townsend Theory is only applicable to such cases
where the distribution of potential between the electrodes is
that given by the geometrical disposition of the electrodes and
is not influenced or altered by the charges of the electrons
and ions themselves. Further as regards the Hoist and Oostpr-
huis Theory it is necessary that the currents passing due to
the ionising factors, shall cause no appreciable excitation of
the gas or production of metastable states.

In circumstances where threshold currents of appreciable
magnitude are maintained through a discharge tube, the
potential required to initiate a self-sustained electrical dis-
charge will be lowered by an amount depending upon the form
of the threshold-current-voltage curve (see Chapt, 2) and this
curve is itself determined by the geometrical disposition and
form of the electrodes, the nature and pressure of the gas,
etc,, etc,. The discharge will become self-sustained as soon as
the voltage across the tube has risen to the value of the
potential on the threshold current curve corresponding to the
threshold current passing through the tube at that instant.
In other words the discharge becomes self-sustained immedia-
tely the threshold current ciu-ve is intersected. Under such
circumstances the function 0 (V, p.) of equations (13) to (15)
is dependent upon the distribution of space charge between
the electrodes and the presence of ionised and excited states,
and becomes of a complicated nature that we may represent
by Cp (V, p, i.), where i is the current flowmg through the tube.
The number of positive ions arriving at the cathode is conse-
quently a function of the actual current flowing. To obtain
the conditions for the initiation of the spark or glow discharge
we must introduce the function (p (V, p, i.) into equations (13),
(14), and (15),

Equation (14) defines the threshold current characteristic
and the portion falling beneath it, which corresponds to an
un-self-sustained electric discharge (see Chapt 2), Unfortun-
ately nothing is known of the function 0 (V, p, i,) except that
it is given empirically by the corona characteristic, which it
explains physically.

In a recent paper Huxley has considered the Photoelectric
Theory of Sparking potentials and concludes that it cannot
be satisfactory. We may therefore consider the objections
raised.

In paragraph (3) of Huxley\'s paper the discharge between
a wire and coaxial cylinder is considered. It is stated that
quot;it is found that the critical force Xj, at the surface of the
wire necessary to initiate the discharge, is independent of the
diameter of the outer cylinder provided the latter exceeds a
certain value. Experimenters who have studied the phenomena
are in general agreement on this pointquot;. In the opinion of the
present writer the experiments on which this generalisation is
based have been performed over a very limited range of
conditions. Nevertheless the conclusion may be accepted as
approximately correct, and Huxley deduces from it quite
rightly that the function r. in the Photoelectric Theory of
Sparking Potentials should be independent of the force at the
negative electrode for this case, (This does not imply however
that r is independent of the pressure of the gas,)

In Table VII are given some results of Watson quoted by
Townsend t) in connection with the above generalisation
relative to the electric force Xj, at the surface of the wire.
It is seen that the ratio of the force X, at the surface of the
outer cylinder to the pressure of the gas p, is of the order
of one volt per cm. per mm. The gas for the case considered
was air.

Huxley then considers the sparking potential for parallel
plate electrodes and concludes that in this case there is a
large variation of y. with the value of X, This appears to bc
contradictory to the conclusion, arrived at for the case of
cylindrical electrodes that rwas constant. This however is not
necessarily the case. We sec from the Table given by Huxley
for air, that X/P for the cases considered is between 131 and
440 volts per cm. per mm.. In the previous case we saw that

E, A. Watson, The Electrician, 11, Feb. (1910).
Cylinder and wire. Outer cylinder diameter 20 cms. Air,

C is the least value of the radius of the outer cylinder for
which the condition relative to the critical field for starting
a discharge, holds.

cp also Townsend and Edmunds, Phil, Mag., 27, 793 (1914),
quot;Discharge of Electricity from cylinders and pointsquot;. quot;In the
results used in the following discussion XJp (that is X/p in
the above results) was always less than 40quot;.

this value was of the order of one volt. These differences for
X/P in the two cases are striking, and it would appear that
the velocity with which the positive ions impinge upon the
cathode is much greater for the case of the plane parallel
electrodes than for the cylindrical ones considered. We are
thus led to infer that y (for a given pressure) may be constant
provided the field at the surface of the cathode is not above
a certain value, or, in other words provided that the positive
ions do not strike against the electrode with too large a
velocity. It can be assumed, as was rendered extremely
plausible by the considerations developed in Chapter 5, that
as the velocity of the positive ions increases so does the
function y. It is also reasonable to suppose that the photo-
electric emission will increase with increase of the surface
field. It is further stated that the photoelectric theory is
unable to explain the large difference between the force at
the surface of a wire, required to start a negative discharge.
This statement cannot be accepted. The electric field at the
cathode is entirely different in the two cases and the cir-
cumstances of the genesis of the electrons from the cathode
and the magnification by collision both lead to a difference
in the required direction.

In paragraph (5) of the paper Dubois\' results 2) are men-
tioned and apparently assumed to be in contradiction to
those of the present writer. This is most difficult to under-
stand, If we do assume that Dubois\' results are due to saline
impurities on the electrodes it is necessary to give some
explanation of their action. An unmodified Townsend Theory
fails entirely to give such an explanation, indeed it implies
that they are without effect at all. On the other hand, if the
effect is due quot;to the action of positive ions in causing elec-
trons to be set free, by bombardment from impurities on the
surfacequot;, then we are immediately introducing a foreign
hypothesis on top of the original one, and since ordinary
electrodes exhibit large variations in the value of the spar-
king potentials the effect attributed to the new action be-

comes of large importance and the Townsend Theory must
be modified (as for instance Dubois himself has done) to
account for it.

Now it is well known that alkali metal impurities very
considerably increase the photoelectric emmissivity of a
metal surface and this strongly suggests that the correct
explanation is afforded by the Photoelectric Theory of Spar-
king Potentials.

It is further stated that quot;Dubois\' conclusions indicate there-
fore, that with ordinary metal electrodes the emission from
the cathode is negligible compared with the action of the

If Dubois\' conclusions are correct and if the action is pro-
duced by the change of the photoelectric function r, there
appears to be no reason why y should not be approximately
the same for ordinary metals for the Schumann radiation
which is the radiation chiefly responsible for the photoelec-
tric emission, for, according to Compton and Richardson, the
photoelectric sensitiveness is supposed to be the same for
all metals, but for the electronegative metals the curve is
shifted bodily towards the region of short wavelengths. It is
obvious then that according to this idea the photoelectric
emissivity of metals, whilst differing enormously for visible
light (near the threshold frequency) may have no great per-
centage difference in the region of shorter wavelengths, that
is in the Schumann region.

Again it is quite conceivable that ordinary metal electrodes
formed in the same way may acquire surface films in the
discharge of similar nature and yield similar results.

In conclusion it may be mentioned that no alternative ex-
planation of the results obtained on pure electrodes and
gases, and on the surface electrode phenomena is given.

It has been found possible to complete experiments upon
the concomitant measurement of igt;f, and y, for the case of
helium.

A tube of the form described on page 12 was utilised. It
was provided additionally with a side-tube containing a four
electrode hot wire discharge tube device, (so that the cathode
of the discharge tube could be irradiated by radiation coming
through a window in the box.) and with a side-tube containing
activated charcoal which could be put in liquid air to absorb
all traces of active gases. The apparatus was thoroughly
outgassed etc.. Carefully purified helium was introduced to
pressures up to about two mms.

It may be assumed that the radiation proceeding from the
low potential discharge box is of approximately the same
character as that arising from the neutralisation of the positive
ions. This radiation falls upon the discharge tube cathode and
produces a photoelectric effect P, that is measurcable by a
sensitive galvanometer. It is evident that P will be a propor-
tional measure of y.

The electrodes of the four electrode device ctc. were charged
to suitable potentials so that no electrons or positive ions were
collected by the discharge tube cathode. It was determined
that the current measured between the cathode and anode of
the discharge tube on irradiation, exhibited all the characte-
ristics of photoelectric currents.

On performing experiments upon the variation of v^ with
progressive purifying of the helium by the cooled charcoal,
remarkable results were obtained. There occurred initially the
well known rapid decrease of the values of the sparking
potential until a minimum value was attained. As purification
was continued, however, a slow rise in sparking potential took
place until a value of anything from 30 to 600 volts (according
to gas pressure etc..) higher than the minimum was attained.

Introduction of new helium to make up for the pressure lost
by the cooling, showed definitely that the effects did not
proceed from the pressvire changes.

Concomitant measurements of P showed a fall from a
maximum at the minimum sparking potential to a minimum
at the final higher sparkmg potential.

After considering a number of explanations of the pheno-
mena, the following was adopted: —

(1)nbsp;The helium rapidly becomes piure, so that only slight
traces of foreign gases remain, a fall of y occurs to a minimum
and the properties of the gas then remain almost constant,

(2)nbsp;The gas layer on the surface of the cathode tmdergoes
slow change, probably by evaporation of surface gas molecules
into the helium and final disappearance in the charcoal. This
slow change of the cathode surface diminishes progressively
its photoelectric emissivity (or its capacity for emitting elec-
trons.) and increase in v^. occurs until the modification of the
cathode has attained equilibrium tmder the existing conditions.

It was determined, for various pressures, that throughout
this region the graphs showing the relation between the
corresponding values of y^ and P were smooth curves, and the
relation between v^, and log, 1/P was cither linear or of slightly
curved form. These findings are in agreement with equations
(15) and (16), At the same time the values of ^ calculated
from the relation of equation (16) agreed, within the limits
of experimental error, with previously determined experi-
mental values (for the higher pressures). We may conse-
quently conclude that the results give strong evidence in favour
of the Photoelectric Theory of Sparking Potentials.

My thanks are due to Mr, v, Hasselt who assisted with much
of the observational work in this experimental verification,

1) Townsend* quot;Electricity in Gasesquot;; Townsend, Hdb. der Radiologie,
i. p. 1020, Hdb. der Physik. xiv, 1927,

5)nbsp;Taylor, Proc. Roy. Soc. A. 114. p. 73 (1927): Phil. Mag. 3, p. 753
(1927).

Under dynamical conditions the sparking potential may
differ considerably from the normal static value. It is only
proposed to touch upon the most important points in this
Chapter. For the more detailed results the works given in
the bibliography at the end of the Chapter are referred to t).

The case of a single flash not preceeded by other flashes,
has been treated by the writer in collaboration with Stephen-
son 2), and more recently the subject has been investigated
by Clarkson 3),

Under such circumstances, provided the threshold cur-
rents passing throught the tube or spark gap, are not large
enough to cause appreciable changes in the sparking poten-
tial, the observed dynamic sparking potential may be con-
siderably greater than the normal statical value, and it has
been shown that the divergence of the sparking potential is
brought about by a time lag in production of discharge be-
hind the voltage tending to produce it 4), If the function lt;$(t),
gives the variation of potential with time, across the discharge
tube terminals, and r, is the time lag in production of dischar-
ge after the time t^., at which the voltage across the tube is
of the value v^, the static sparking potential, then the dynamic
sparking potential is given by the relation,

Alternatively if (V) is the current flowing into the dischar-
ge tube system (Capacity C) for a given voltage V, across
the electrodes, before discharge occurs then

For cases in which C, is large, as for example when a capa-
city of some microfarads is placed in parallel with the tube,
and r. is not very great, we have that V^ is of approximately
the value of the statical sparking potential v^ For small
values of C, however, and for large values of\' r, V^ may
diverge very considerably from the normal statical value.

The case described above is important from the point of
view of its technical applications to sparking gaps, sparking
plugs etc, 5), Equations (17) and (18) are easily adaptible to
the requirements of any particular case.

There are several theories on the subject, none of which
are entirely satisfactory. The most usual theory is that the
lags are brought about by a lack of electrons or ions (within
the gas) to produce a discharge by ionisation by collision
(see Chapter 2, quot;The initiation of the dischargequot;). This
theory is supported by the results of Zuber6), According to
Zuber the time lag is determined by chance, and obeys a
probability law, Zeleny?) attributes the lags in the case of
discharges from points, to the surface layers of gas on the
electrodes. Peek 8 explains the lag as being the time of build
up of the corona discharge. Pedersen9) in work on lags of
very small duration (of the order of 10quot;^ sees.) concludes
that, for clean electrodes and dry gases, the lag is constant,
independent of the electrode distance and of the intensity of
the irradiation.

Time lags of long duration (half hour or more) may be ob-
served if the potential applied to the discharge tube is just
of the static sparking potential. The time of lag is variable,
but the average time decreases with increase of the applied
potential and finally becomes very small 4).

cathode (also irradiation of the gas in some cases) reduces
considerably the time of lag, and if the ionising factors arc
sufficiently great the lag becomes inappreciably small 10), ¹)

(1).nbsp;Irradiation of the cathode and (or) gas of the discharge
tube will ensure the presence of electrons or ions in the gas.
Consequently the lags due to the lack of ionisation required
to build up discharge will be suppressed or reduced to small
values. Certainly in the majority of cases lags of duration of
the order of several minutes are suppressed in this manner 4),
This is not invariably the case however, indeed in certain
tubes used by the present writer, no appreciable lag was
observed when the potential across the tube was of the static
sparking value, either in darkness or in the presence of ioni-
sing factors, but lags of considerable duration were obser-
ved either in the dark or in the presence of ionising factors
when the tube became polarised. Indeed, it was found that
with such discharges in rare gases, the time lag appeared to
be connected essentially with a state of polarisation of the
electrodes.

Later work by Clarkson 3) has disclosed the extremely
interesting fact that lags of considerable duration (of the
order of a hundreth of a sec,) may occur in such tubes even
when relatively large threshold chrrents are passing, Clark-
son is further of opinion that the lags observed in his experi-
ments are to be referred to the electrode surface effects on
the cathode.

1nbsp; The smallest possible lag is of course given by the time required
for the actual building up of the discharge, that is to say the time of
relaxation of the gas. This time is usually extremely small, of the order
of 10-r® sees.

layers on the surface and will consequently modify any iag
phenomena which is caused by such effects.

When a potential equal to the normal statical sparking
potential is placed across an unpolarised discharge tube,
discharge will occur. The passage of the discharge is accomp-
anied in many cases (see Chapter 3. quot;On polarisation Pheno-
menaquot;) by a charging up positively of the surface layer of
the cathode (occluded gas layer). This action sets up an elec-
trical double layer which will disappear with time, the time
of disappearance varying according to the type of layer. The
potential across the surface layer will depend, as we saw
previously, upon the length of time for which the discharge
has been running etc. When a potential is again placed across
the discharge tube electrodes this polarised layer will modify
the conditions under which discharge occurs.

In the first place if the potential across the gas layer is
p volts, then if V is the potential across the tube, the actual
effective potential across cathode to anode is (V-p) volts.
The experimental results indicate that p is fairly small so
that this effect need no be considered in any detail.

Secondly the charge of the layer will introduce a decrease
of the photoelectric emissivity of the cathode, for the elec-
trons must traverse the charged layer of gas and the work
of extraction is consequently increased by the amount cor-
responding to the potential p. volts. According to the photo-
electric theory of the sparking potentials this change of
photoelectric emissivity will be accompanied by a conside-
rable increase in the sparking potential.

Thirdly, if the surface layers on the cathode are of con-
siderable thickness they may introduce a considerable in-
crease in the resistance of the discharge tube and decrease
the possible energy transference.

On these views a considerable part of the lag in the
discharge behind the voltage tending to produce it, may be
caused by the surface layer action.

the time taken for the disappearance or partial disappearance
of the charge of the layer, in others the lag may proceed from
the setting up of a layer in the initial stages of the discharges.
Such mechanisms would introduce time lags of variable dura-
tion according to the previous electrical treatment of the
tube.

It is not possible to state definitely of course that all time
lags in the discharge are attributable to such a mechanism,
indeed it is somewhat improbable, but that such an action
enters into the lag phenomena is proved definitely by the
writer\'s work, and the more recent work of Clarkson,

We see at once that the action of ionising lactors may in-
troduce large variations in the lag. The effects of irradiation
may be of several kinds.

The photoelectric action of radiation incident upon the
cathode will cause the emission of electrons, which will
hasten the neutralisation of the charged surface layer. The
efficiency of the radiation will increase with the intensity
and frequency. Further, if the quantum of the incident radia-
tion is sufficiently great the photoelectrons will be ejected
with sufficient velocity to penetrate the polarisation layer
and supply electrons for the building up of the discharge.
Both these actions will entail a diminution of the time of lag.

(3) Irradiating the cathode or (and) gas will set up a
threshold current which will lower the sparking potential.

This effect was considered fully in Chapter 2, The magni-
tude will depend upon the intensity and frequency of the
radiation, the form and geometrical disposition of the elec-
trodes, the nature and pressure of the gas, etc.

In any actual case the sparking potential under dynamical
conditions will be determined by the two coexistent and coin-
cident phenomena, the lag in discharge production, and the
threshold current traversing the tube. It is evident therefore

that the dynamical sparking potential may be greater than,
equal to, or less than the normal statical sparking potential
according to circumstances.

For the case of continuous flashing in discharge tubes there
is much experimental evidence in favour of the above hypo-
thesis. In some recent work on the lag phenomena in flashing
of certain types of discharges, Clarkson arrives at the con-
clusion that the sparking potential changes arise from the
two effects of time lag in the production of discharge, and the
persistence of threshold currents during the quot;darkquot; period 3).

It should be stressed here that persistence of the metas-
table states may also be a considerable factor in determining
the sparking potential variations observed in such cases.

The working of the above described independent effects
throws considerable light upon the divergent results obtai-
ned by numerous observers working under different con-
ditions.

1)nbsp;The literature on this subject is very extensive. See e.g. references,
8, 9, 10, Chapter 1, also Valle, Phys. Zeits. 27. p, 473 (1926; Penning, ibid.
27, p. 187 (1926); Schallreuter, quot;Über Schwingungscrscheinungcn in Ent-
ladungsrohrenquot;, Sammlung Viewcg, 66; Taylor, Clarkson and Stephenson,
Journ. Scient. Inslrs. 11. p. 154 (1925),

5)nbsp;Peak. quot;Dielectric Phenomenaquot;; Campbell, Phit. Mag. xxxviii. p. 214
(1919); Wynn-WilUams, ibid. 1, Feb. (1926); Morgan, ibid. 4. p. 91 (1927).

10)nbsp;Sec e.g. Ziiber, loc. cit.; Taylor, Clarkson and Stephenson, loc. cit-:
Taylor and Sayce, Phil. Mag. 1. p. 918 (1925),

\'V

• • •■nbsp;.M .n
- ; V;m. W • .{rflt;-?- i.,,; gt; .|h¥ .V4H (Ç,.^ \' ;.

. ■ •• ■ quot; .nbsp;«I ,4tquot;\'gt; .ÙÂ\'T II
■ .-\'T-Î^iî ..m ..q .ärH iéüïA^-vl»

The objection of Salet to the Theory of Solar
Prominences of Julius does not necessarily disprove the
theory.

The statement „finally it seems quite likely that,
when the total characteristic (including the parts with a
negative slope) is taken into account, the well-known
vibration of a neon-tube connected to a resistance and
condenser in shunt may be similarly treated under the
heading of relaxation-oscillationsquot; is not correct.

Dauvillier\'s experiments on the pressure effects
and radiations in rare gas discharges are not entirely
satisfactory.

The term „factor of merit of a galvanometerquot; as
used in English scientific literature is misleading and in
many cases has little significance.

It is possible to measure under precise conditions,
time lags in the production of spark or glow discharges,
and such works as those of Campbell, Zuber and Pedersen,
have been done under conditions unsuitable for the
development of the physical theory of the subject and
are needlessly complicated.

• • The policy adopted by the special Committee for
thequot; League of Nations Advisory and Technical Committee
for Communications and Transit, in the consideration of
calendar reform is unscientific.

The Coolidge type of x-ray tube can only be
applied to the examination of the characteristic radiation
of targets if great precautions are taken.

The statement that „the prestige of physics has
exerted a harmful influence on the study of psychologyquot;
is not correct.

The work of Boomer on the formation of helium
compounds in the electric discharge is not entirely satis-
factory.

The application of the „exchangequot; system to the
teaching profession would be attended by beneficial
results.

eiirf g\'jia^HqVio \'üXiiJH\'^iq- ari».,: iiirü .^narrf^tciJ?.\'sirlT r \'
.. ■ \'• ,nbsp;.nbsp;\'nbsp;.loaitooquot;ipii al

quot;rnjiilo/i hgt;. iioiXßmioi 3tii rii) ;îôim)()-8 lo gt;l!óv/ a^flT..
-r-.i^ft^.^Ia\'iiin\'i k)n rit ojnr.rb«ib ^yiuiôb ;lt;i,ijî fii ^{biujlt;gt;q[{iO\'j.

.oîi) oi niyJ^YS \'\'\'gt;;;ùii;fi\'/Xv,. î5([j fo [loiinoiiqqji i^riT . , \'
Ijibibnyrl \'{flnbsp;od. biii jw n()igt;i;^,i)toiq ti(nijrbiwJ.

- ; -a;

. , ■\';■ ■ivï\'V- ■

■U\'.\'.î-

wv-Sgii

Polarisation in volts. A. cathode.	No. of flashes.	Polarisation in volts. B cathode.	No. of flashes.
5	4	8	5 or 6
6.5	6 or 7	10	4 or 5
7.5	6 or 7	10	6
5.7	6	8	6
8	6	6.5	8
9.5	10	11	6
Averages... 7	6.5	9	6

Sparking Potentials of first discharges.	Treatment of Tube. (All heatings took place in the gas.)
S Cathode.	W Cathode.
Volts.	Volts.
275	251	W heated strongly a few minutes.
244.5	245	S and W heated for a few minutes.
264	247	S and W heated strongly for 10 minutes.
247	255	W heated very strongly for a few minutes.
267	255	S heated strongly a few minutes.
251	251	W heated for a few minutes.

S Cathode.	W Cathode,	Treatment of Tube,
Volts,	Volts.
122	106	Both electrodes quot;sodiumated.quot;
136	214	W heated at dull red heat for 1.5 minutes.
142	98	S heated for an instant to dull red.
152	95	S heated dull red for a second or two.
157	93	Do.
183	101	S heated dull red for 10 scconds.
230	108	Do.
254	110	Do.
254	111	Do,
253	112	S heated dull red for 20 seconds.

Pressure p mms.	Diameter of wire 2a cms.	X. kilovlts.	X kilovlts.	X/P (volts.)
760	0.1	75	0.375	0.49
560	0.136	55	0.375	0.67
360	0.211	34	0.358	0.99
760	0.2	61	0.61	0.8
560	0.272	44.5	0.605	1.08
360	0.422	28.5	0.6	1.67
760	0.5	46.5	1.16	1.52
560	0.68	35.0	1.19	2.12
360	1.055	22.0	1.16	3.22

p. mms.	radius wire cms.	C. cms.	X, kilovlts.	Xc kilovlts.	Xc/P volts per cm. per mm. pres.
760	0.5	0.66	40.0	30.3	40
360		0.82	23.0	14.0	39
108	.,	1.12	9.45	4.21	39
25.2		1.73	3.4	0.98	39

..... . .. • • ;gt; .■., O-. .,;,; ,