the electredes only if the avalanche is 
sufficient to overcome the strong nega- 
tive space charge present near the wire. 
This explains the immunity of the spark 
counter to beta and gamma radiations, 
which produce a much weaker ava- 
lanche than does a heavier ionizing par- 
ticle like the alpha particle. 
On further increase of Vz, the a 
priori probability for an avalanche to 
produce a spark breakdown, or a priori 
sensitivity, increases, causing an in- 
creased number of alpha particles from 
the source to be counted. It may hap- 
pen that when V, has increased beyond 
a certain limit, the corona-space-charge 
effect, which also increases with Va, is 
also very high. This might lead to a 
reduction in a priori sensitivity, since 
the limited specific ionization of some 
a 
D 
> 
? 
Extrapolated 49°C — 
Shift 
nD 
Normalized Counting Rate{ 10°cpm) 
fo) 1.0 20 30. 
_ Distance of Source from Wire (cm) 
FIG. 6. Solid curves are counts ob- 
served at 20° C and 49° C as Po?!0 
source approaches counter. Dotted 
curve is shifted version of 20° curve 
of the particles may not be sufficient to 
overcome the space charge and may 
thereby fail to produce a breakdown. 
We can then expect a slight fall in the 
counting rate although J, is rising. 
For a heavily ionizing particle like an 
alpha the specific ionization in the 
counting space depends somewhat upon 
the distance of the source from the 
counter. Sparking breakdown requires 
a critical concentration of ionization in 
the sensitive space. The variation of 
the source distance or the direction of 
incidence of the alpha particle may, 
therefore, cause a variation in the a 
priori sensitivity, giving the spark coun- 
ter strong directional properties. 
Let us now examine the effect of R 
and V, on Vg and the counting rate of 
the counter. It can be seen from theo- 
retical consideration that for a given 
value of R, an approximately linear 
quiescent-current characteristic exists 
within a certain range of V,; that is, as 
VY, increases, J also increases linearly. 
The magnitude of J is, however, deter- 
mined by both the value of R and the 
dynamic resistance of the space-charge 
sheath. On increasing R for a given 
V., the current J may therefore de- 
crease nonlinearly with R. We may 
thus represent J as 
1 =k(R)V. 
where k(R) is the constant slope of the 
I vs V, characteristic. The quantity 
k(R) is a constant for a given R. Sub- 
stituting in our earlier relation, one gets 
Vou = Vall — k(R)R] 
From this equation it follows that for 
values of k(R) < 1/R, Vor will increase 
with V,. This will be true only for cer- 
tain values of R. For some other 
(larger) values of R, k(R) may exceed 
1/R, giving a reduction in Veg with 
increasing V,. If the electrode spacing 
is small, k(R) can exceed 1/R even for 
moderate values of R. The conse- 
quence of this effect will be a decreasing 
a priori sensitivity and a decreasing 
counting rate with increasing V,. 
For a given electrode spacing there is 
therefore a critical resistor value R, 
above which the counting rate will de- 
crease with increasing V.. Further, 
R, will decrease with decreasing elec- 
trode spacing. 
Experimental results on counting and 
quiescent-current characteristics, the 
number-distance curve for polonium 
alphas, the temperature coefficient of 
counting, and the detection of neutrons 
can be explained by the counter mecha- 
nism outlined. Some of the apparently 
conflicting results of earlier workers 
can also be accounted for. 
Counting Characteristics 
A polonium source on a 1-cm-diame- 
ter thin platinum disk was placed fac- 
ing the counter on a line normal to the 
plate. The variation of the counting 
rate with applied voltage was observed. 
Figure 3 shows the result. It differs 
from what earlier workers have ob- 
served (3-6). 
The difference between our result and 
those of Connor and Payne can be 
understood as follows. Alpha rays 
from our extended polonium source en- 
ter the counter at many different angles. 
Since with increase in. V,, a priori sensi- 
tivity gradually increases, some of the 
oblique rays that could not produce 
sparking breakdown at lower voltages 
are counted at higher voltages. This 
gives rise to the small positive slope of 
our counting characteristic. If the 
source is highly collimated, the direc- 
tional effect on the counting rate will 
be practically absent, but the effect of 
increasing space charge and that of the 
external resistor as explained previ- 
ously will still remain. These may re- 
sult in a positive, zero, or negative slope 
of tle counting characteristic, depend- 
ing on the particular circumstances of 
the experiment. 
It appears to us that the conditions 
and geometry of Connor’s counting ar- 
rangement (4) were similar to ours ex- 
cept that he used a smaller separation 
between the electrodes. The latter 
might be responsible for a strong space- 
charge effect at higher voltages. Also, 
the external resistor used in his case 
might exceed the critical value R, for 
his small electrode spacing. These two 
effects might combine to give a slow 
decrease in the counting rate with in- 
creasing voltage. Exact reasons can 
not be assigned without knowing the 
quiescent-current characteristics ob- 
tained in Connor’s experiment. The 
nearly constant rate observed by Payne 
(6) can also be similarly understood in 
a general way. 
To investigate the effect of source 
collimation we observed counting char- 
acteristics with two different collimat- 
ing tubes of 1.4- and 0.6-mm diameters 
mounted over the polonium source. 
Figure 4 shows the results. It is clear 
from the three plots that the slope of 
the counting characteristic progres- 
sively decreases on putting the colli- 
mators in front of an extended source 
and even becomes negative for a highly 
collimated source. Clearly the ab- 
sence of oblique rays for a collimated 
source reduces the slope. The increas- 
ing Ve can even make the slope nega- 
tive, as is probably the case for the 
lowest curve. The middle curve repre- 
sents perhaps an intermediate case 
with limited collimation. The ob- 
served negative slope of the high-colli- 
mation curve is not the effect of the 
external resistor in our case, as will be 
clear from discussion further on, but 
should be attributed to the collimation 
and space-charge effects. 
When examined on an oscilloscope, 
the counter pulses were found to be of 
more or less the same magnitude for a 
fixed applied voltage and the pulse- 
39 
