248 H. A. BETHE AND C. L.

CRITCHFIELD

m a t t e r is approximately the same as t h a t of the
showers observed a t sea level. T h e general
features of this conclusion are in agreement with
the results of an earlier investigation 5 on shower
production under thick layers of various ma-
terials. W e observed transition effects which
indicated t h a t the showers were electronic in
character. W e suggested t h a t such showers might
have their origin in the generation of shower
producing radiation (possibly secondary elec-
trons and photons) by the penetrating com-
ponent. I t also follows from these transition
curves t h a t some of such secondary particles
must have considerable energy.
We wish to t h a n k Mr. J. Q. Gilkey of Marion,
N o r t h Carolina for his kindness in making Lin-
ville Caverns available for these measurements
and to acknowledge the interest of Professor L.
W. Nordheim and his helpful suggestions in
our work.

AUGUST 15, 1931 PHYSICAL REVIEW VOLUME 54

The Formation of Deuterons by Proton Combination

H. A. BETHE, Cornell University, Ithaca, N. Y.
AND

C. L. CRITCHFIELD, George Washington University, Washington, D. C.

(Received June 23, 1938)

The probability of the astrophysically important reaction H + H = D + e+ is calculated. For

the probability of positron emission, Fermi's theory is used. The penetration of the protons
through their mutual potential barrier, and the transition probability to the deuteron state,
can be calculated exactly, using the known interaction between two protons. The energy
evolution due to the reaction is about 2 ergs per gram per second under the conditions prevailing
at the center of the sun (density 80, hydrogen content 35 percent by weight, temperature 2 • 107
degrees). This is almost but not quite sufficient to explain the observed average energy evolu-
tion of the sun (2 ergs/g sec.) because only a small part of the sun has high temperature and
density. The reaction rate depends on the temperature approximately as T3-5 for temperatures
around 2 • 107 degrees.

§1. INTRODUCTION Weizsacker. 1 However, there seems to be a

general belief t h a t reaction (1) is too rare to
I T seems now generally accepted t h a t the
energy production in most stars is due to
account for a n y appreciable fraction of the
energy production in stars and t h a t it can serve
nuclear reactions involving light elements. Of all
only to start the evolution of elements in a star
the elements, hydrogen is favored by its large
which will then be carried on by other, more
abundance, b y its large internal energy which
probable, processes. I t is the purpose of this
makes a considerable energy evolution possible,
paper to show t h a t this belief is unfounded b u t
and by its small charge and mass which enable it
t h a t reaction (1) gives an energy evolution of the
t o penetrate easily through nuclear potential
correct order of magnitude for the sun.
barriers. Again, of all reactions involving hydro-
On the other hand, we do not w a n t to imply
gen, the most primitive is the combination of two
t h a t reaction (1) is the only important source of
protons to form a deuteron, with positron
energy. An analysis of all possible nuclear
emission:
reactions with light elements 2 shows t h a t the
H + H = D + e+. (1)
capture of protons b y carbon a n d nitrogen will
In fact, this reaction m u s t stand in the beginning also play an i m p o r t a n t role. I t is more i m p o r t a n t
of a n y building u p of chemical elements; it has
*v. Weizsacker, Physik. Zeits. 38, 176 (1937).
already been discussed in this connection by v. 2
Bethe, to appear shortly in the Physical Review.
 FORMATION
t h a n (1) for heavy stars, less i m p o r t a n t for light
ones and a b o u t equally i m p o r t a n t for the sun.
In calculating the energy evolution from reac-
tion (1), it m u s t be considered t h a t (1) is followed
b y a n u m b e r of other reactions which are all
" f a s t " in comparison with (1) because they
involve the emission of radiation or of heavy
particles rather t h a n of /3-rays. 3 T h e deuterons
formed in reaction (1) will first capture another
proton, with 7-ray emission:
D + H = He3+7- (2)
T h e ratio of the rates of reactions (2) and (1) is
a b o u t 10 18 which means t h a t , in equilibrium,
there will be a b o u t one deuteron for 1018 protons.
This makes collisions between two deuterons
very improbable and t h u s excludes a n y ap-
preciable production of neutrons b y such
collisions.
T h e fate of the He 3 formed in (2) depends to
some extent on the question whether this nucleus
or H 3 is more stable. T h e most recent determi-
nation 4 of the energy evolution in the reaction
H 2 + H 2 = H e 3 + n 1 seems to show t h a t He 3 is more
stable while older determinations 5 gave the
opposite result. T h e processes which occur are :
If He 3 is more s t a b l e :
He3+He4 = Be7+T,
Be 7 = L i 7 + e + , (3)
L i + H = 2He 4 .
7
If H 3 is more s t a b l e :
H e 3 + € - = H3,
H3+H = He 4 + 7 . W
T h e net effect is in both cases the same, viz. the
combination of four protons a n d two electrons to
form an ex-particle. (The emission of a positron is
equivalent to the consumption of an electron
since the positron will ultimately annihilate an
electron.) One a-particle is formed for each
process (1). T h e energy produced per process
(1) is therefore equal to the difference in weight
between four hydrogen a t o m s and one helium
a t o m , viz. (cf. reference 5, 6) (2 = 4-1.008 13
3
4
Cf. Bethe, Rev. Mod. Phys. 9, 188, § 76C (1937).
5
Bonner, Phys. Rev. 53, 711 (1938).
See Livingston and Bethe, Rev. Mod. Phys. 9, 373
(1937).
6
Bainbridge, Phys. Rev. 53, 922A (1938).
OF DEUTERONS 249
- 4 . 0 0 3 86 = 0.028 66 mass units, i.e.,
<2 = 4 . 3 - 1 0 - 5 e r g . (5)
While the energy evolution in (3) and (4) is
exactly the same, the processes lead to t h e
formation of different intermediate products
which could be of influence on the building u p of
heavier elements. However, it can be shown 2 t h a t
the building up of elements heavier t h a n He 4 is
negligible under a n y circumstances so t h a t the
question of stability of He 3 or H 3 is actually n o t
important.
In order to calculate the probability of the
proton combination, the following factors have to
be considered:
(1) T h e probability of collision of two protons
which involves the penetrability of their m u t u a l
potential barrier. This can be calculated very
accurately since the force between two protons is
very well known from scattering experiments. 7
(2) T h e probability of emission of a positron
during the collision. This involves a theory of the
0-decay. All available experimental evidence
points to the conclusion t h a t the original Fermi
theory gives good agreement with the experi-
mental dependence of lifetime on energy.
(3) T h e energy distribution of the protons in
the star which is given b y the Boltzmann law.
§2. T H E P R O B A B I L I T Y O F P O S I T R O N E M I S S I O N
According to Fermi's theory, the probability of
/3-emission (per second) is
P=gf(W)\G\\ (6)
Here G is the matrix element of the nuclear
transition,
G=f^fdrf (7)
\{/i and \pf the initial and final state of the
nucleus, except for the substitution of one
neutron b y a proton. / is a function of the energy
W of the emitted ^-particle,
/l • 3 2\
f(W) = (W2-1)*( —W* W2 I
V30 20 15/
1
+-w\og tw+(w2-iy^, (8)
4
7
Tuve, Heydenburg and Hafstad, Phys. Rev. 50, 806
(1936).
250 H. A. BETHE AND
where W is expressed in units of mc2 and includes
the rest mass. A lower limit for the Fermi
constant g can be deduced b y assuming t h a t G = 1
for transformations such a s N 13 —>C l3 +€ + . This
reaction has a half-life of 11.0 minutes and an
upper limit of the ^-spectrum (observed) of a b o u t
1.25 Mev, therefore jS = log 2/660 = 1.05-10~ 3
sec." 1 , W=3.5,f(W) = 11.86 a n d
g = 0.9-10- 4 sec." 1 . (9)
T h e (maximum) energy of the positron emitted
in the combination of two protons follows from
the very accurately known masses of proton and
deuteron, 5 viz.:
T w o p r o t o n s = 2-(1.008 1 3 - 0 . 0 0 0 55) =2.015 16
Deuteron = 2.014 7 3 - 0 . 0 0 0 55 = 2 . 0 1 4 18
Positron energy, including mc2 0.000 98
(mass units), equivalent to 1830-0.000 98 mc2
= 1.80mc 2 .
For W= 1.8, we have/(TF) = 0.132 and therefore,
with (6) and (9),
5
i8=1.2-10- |G|2sec.-1. (10)
A similar value is obtained b y extrapolating
the empirical Sargent curves t o lower energy.
/3 would be larger, if the matrix element G for the
transformation N13—*C13 is less t h a n one. On t h e
other hand, the Konopinski-Uhlenbeck theory
would give a b o u t 10 times less for the decay
constant. However, t h a t theory seems to give too
strong a dependence of decay constant on ener*gy
throughout, and will therefore not be used.
In calculating the matrix element G for the
proton combination (1), i t must be considered
t h a t the wave function of the deuteron is sym-
metric with respect to interchange of space and
spin coordinates of proton and neutron while the
wave function of two protons is antisymmetric. 8
In particular, when t h e two protons are in an 5
state (which is most favorable for their coming
close together), their spins will be antiparallel
(singlet) whereas the ground state of the deuteron
is a triplet S state. T h e transition is therefore
allowed only if the Gamow-Teller form 9 of the
jS-theory is used. This formulation permits a
8
We are indebted to Professor Oppenheimer for drawing
our attention to this point.
9 10
Gamow and Teller, Phys. Rev. 49, 895 (1936).
C. L. CRITCHFIELD
change of direction of the spin of the proton
which transforms into a neutron, and therefore a
change of the total spin b y one unit. Strong
evidence for this theory is found 10 in the short
life of H e 6 ; the ^-transformation He6—>Li6 leads
very probably from a lS t o a ZS state and is
closely analogous to our process 2H—>D. We shall
therefore accept the Gamow-Teller theory in this
paper. Moreover, we shall assume t h a t the matrix
element G in (10) can be calculated simply b y
integrating over the spatial coordinates of the
two particles, i.e., t h a t the summation over spins
which is also implied in G gives unity. This
assumption does not seem serious in view of the
uncertainty in the numerical factor in (10).
If the original Fermi theory were taken instead
of t h e Gamow-Teller theory, t h e transition
H + H = D + e + could occur only b y virtue of the
"small" components of the Fermi interaction, viz.
^heavy particle'^electron-neutrino* oince t n e UiraC
operator «heavy (velocity) is " o d d , " its application
on the ZS wave function of the deuteron ground
state will give a function of SP character. This can
combine with the ZP p a r t of the wave function of
the incident protons. Due to the smallness of the
latter and of t h e a-operator itself, the transition
probability will b e greatly reduced, viz., ap-
proximately b y a factor
(W/I)2(E/Mc2) «10~ 5 ,
where W=1.S mc2 is the energy evolution in the
process, 7 = 4.4 mc2 the binding energy of the
deuteron, £ ^ 5 0 kev the kinetic energy of the
protons and Mc2 = 931 Mev. In this case, then,
the energy evolution in the sun due t o proton
combination would be negligibly small.
§3. T H E C O L L I S I O N C R O S S S E C T I O N
According to the general principles of q u a n t u m
theory, the cross section a for the combination of
two protons of relative velocity v is given by (6),
(7) if we insert for \f/i the wave function of two
protons normalized t o unit incident current.
T h u s we have
*=gf(W)irl\SMdT\\ (11)
where ypa is the wave function of the ground state
of the deuteron and \//p is normalized per unit
Goldhaber, Phys. Rev. (to be published).
 FORMATION OF DEUTERONS 251

density a t infinity. T h e integration goes over t h e of t h e Coulomb equation which behave a s y m p -

space of t h e relative coordinates of t h e two
protons (or proton and neutron, in the deuteron).
If we assume a square potential well of radius
VQ and depth Vo, the normalized wave function of
the deuteron is 11
totically as sin (p — <p0) a n d cos (p — <p0), re-
spectively. As can be seen easily, (15) goes over
asymptotically into (13). K is the phase shift due
to t h e nuclear field, fixed b y t h e condition of
continuity of (14), (15) a t r = r0; we have 1 2

= (J5/r) exp (-(x-xo)) for r>rQ ' . t3LnK = F25/[l-FG5'] (16)

(B/r) sin jux/sin juxo for K r 0
d log F d log w\
with t h e abbreviations ( ——
dp
) .
dp J ro
(16a)

x — r/b) xo = rQ/b> (12a)

For t h e small proton energies concerned, F(TQ)
& = #(il/e)-* = 4.37-10~ 1 3 cm, (12b) is very small (see below), both because of t h e
Coulomb barrier a n d t h e small value of k.
/*=(70-6/6)*f (12C)
Therefore K will be very small (i£ = 0.0017 for
B = (2ir&)-*(l + * o ) - * ( l + M - 2 ) - * . (12d) v = e2/h, i.e. for a relative kinetic energy of
12.5 kev which gives t h e largest contribution to
M is t h e proton mass, e the binding energy of the
the cross section a t a temperature of 5 • 10 7
deuteron, i.e., 2.17 Mev. D e p t h a n d width of t h e
degrees) a n d eiK cos K (cf. (15)) can be replaced
well are related b y t h e condition
by unity. Then t h e factor A in t h e internal
jn cot juxo= — 1. (12e) wave function (14) is
T h e wave function of two protons, normalized F{r,)+G{r,)tznK
ikz A=— . (17)
to unit density, h a s t h e form e a t large dis- k sin (MD)*r0/h
tances. (For considerations of symmetry, see §4.)
I t can be expanded in spherical harmonics and, According to Yost, Wheeler a n d Breit, 12 t h e
a t t h e small velocities prevailing in stars ( ^ 1 0 Coulomb wave functions m a y be written (for
orbital m o m e n t u m L = 0)
kilo volts), only t h e zero order t e r m will be
i m p o r t a n t . I t has, a t large distances, t h e form F=Cp$, G=C~X®, (18)
V^ = sin (p-<p)/p, (13) where
C={2TTri)h~^\ rj = e2/hv (19)
where P = kr, k = Mv/2h (13a)
contains t h e effect of t h e Coulomb barrier, a n d
and <p is a phase (depending logarithmically on $ a n d © are slowly varying functions of r which
p). A t small distances, i.e., inside t h e nuclear have the value 1 for r = 0. Inserting into (16), (15)
potential well, we have we get • •
w A (MD)* tani£=C 2 jb- 0 A, (20)
$p — — = — s i n r, (14) 2
r r h \ = $ f/(l-<j>@f), (20a)
d log F/w\
where D is t h e depth of t h e potential well
between t w o protons, assumed i"square" a n d of
( d log p A o
radius ro, A is a normalization factor. Outside of d log $ d log w\
the well, ypv is a solution of t h e Schroedinger
equation in t h e Coulomb field which m a y be
( l+r
dr
r ) , (20b)
dr / ro
written 1 2 **=C[*(r)+X(r0/r)©(r)] (r>r0), (21)
&, = «** cos K(F+G t a n K)/kr. (15) To
lk=C--[>(ro)+X0(ro)]
Here F is the regular and G the irregular solution r
11
Bethe and Bacher, Rev. Mod. Phys. 8, 110 (1936). sin vr/b
12
Yost, Wheeler and Breit, Phys. Rev. 49, 174 (1936); X _ (r<ro). (22)
J, Terr. Mag. 40, 443 (1935). sin vro/b
252 H. A. BETHE A N D C. L. CRITCHFIELD

In the last expression, we have introduced the A2=l+x0+(2b/a)(l+Xo+ixo2)

abbreviation
+±(2b/ay(l+x0+ixo2+W)+' '', (26b)
1
v=(MD)ih- b=(D/e)i, (22a)
A8 = Xxo{l + ( 2 6 / a ) C l o g ( 2 6 / a ) + 2 7 - l ] A 2
which will be useful for the integration. X repre- -l(2b/ayUl+xQ+ixo2)
sents the effect of the resonance due to the
nuclear potential well. +±(2b/ay(7/3)(l+Xo+W+W)
For slow protons, the wave functions a t small
r are almost independent of the proton energy + • •] + E(26/«)a/.(^o)/(j-l)! (26c)
s=l
and have the form 12
with
$(y)=:r»/1(2y*)
/ s ( # o ) = s ! _ 1 I exp (—(x—Xo))xs log xdx, (26d)
yZ
= 1 +1!2! (23)
2!3! 3!4! / i = l + ( l + x o ) l o g x 0 — £»•( —x 0 )exp(x 0 ), (26e)
0 ( y ) = -2y*Ki(2y*) = l+y(log y+ 2y-1)$ /2 = f + J x o + ( l + x 0 + i ^ o 2 ) log^o
u»+i • /l 1 —Ei(-xo) exp (# 0 ), (26f)
- Z - ; E ( - + ) (23a) 2
?+l!^iU /+! /3=(ll/6H(5/6)x0+i*o
with + ( l + x 0 + J x o 2 + i x 0 3 ) log^o
2 2
y = 2r/a, a = 2h /Me = 5.75-10~™ cm, (23b) — £<(—Xo) exp (x 0 ). (26g)
7 = 0.577- • • (Euler's c o n s t a n t ) .
Table I gives the numerical results for two
T h e deviations from (23), (23a) for rj = 1 are of
values of the radius r 0 of the potential well,
the order of a percent. viz. e2/mc2 and e2/2 mc2. T h e depth D of the
Collecting our formulae, we m a y now write for proton-proton well was taken from Breit, Condon
the cross section (11) and Present. 1 3 All other quantities were calcu-
gf(W) (4^)2 lated from the formulae in the text. I t is seen
<r = - C2 A2, (24) t h a t the contribution of the inside of the poten-
V 2irb tial well, Ai, is rather small even for the larger r0,
where which shows t h a t the result will not depend
sensitively on the shape of the potential well.
(l+x0)4(l+M-2)iA = A1+A2+A3, (25)
TABLE I. Numerical results for two values of the radius.
xoOW+xeo-o)] r*\ , _, //ip x
Ai = I sin jnic sin ^ m , (25a) ro=e2/mci ro=e*/2 mc2
sin JJLXO sin J>XO «^o
Xo 0.645 0.322
V0 (Mev) 20.9 66.5
A 2 = I xdx exp ( — (x — xo))$(2bx/a), (25b) D (Mev) 10.3 47.0
A* 2.94 5.45
V 2.18 4.65
(rd log w/dr) 0.236 0.110
A3 = Xx0 I d# exp ( — (x — xo))®(2bx/a). (25c) *(n>) 1.050 1.025
O(ro) 0.769 0.854
(rd log &/dr) 0.050 0.025
f 0.814 0.915
T h e integrations are elementary and give X 2.63 4.80
A! 0.689 0.277
i*o2l>(ro)+X0(ro)] A2 1.949 1.547
Ai = A3 1.205 1.030
(l+Xo)(l+M- 2 ) 1.835 1.367
sin /ZJCO sin VXQ A2 2.84 2.44
A 8.08 5.93
sin (^ — V)XQ sin (/X+J0#O"1
X (26a)
13
(/i — *>)x0 -V)XQ
(V-\-V)XQ J Breit, Condon and Present, Phys. Rev. 50, 825 (1936).
 FORMATION OF DEUTERONS 253

Furthermore, the contribution A3 which is due to TABLE II. Energy evolution

2
due to zproton combination, for
pCH =10 gjcm ,
resonance is smaller t h a n A2 which would be
present even if there were only the Coulomb t(million degrees) 5 10 15 20 30 40 50 100
field between the two protons. T h e final result T 19.8 15.70 13.72 12.56 10.89 9.90 9.18 7.28
e(ergs/g sec.) 0.0040 0.15 0.76 2.2 9.1 20 36 150
for A2 increases somewhat with increasing radius
of the well. For rj = l, a numerical calculation
gave A2 = 8.8 instead of 8.1. factor j for the a priori probability of opposite
spin, there remains a factor J.
§4. SYMMETRY, S T A T I S T I C S , E T C . T h e total number of processes per cm 3 per sec.
becomes thus
If N is the n u m b e r of protons per cm 3 , the
probability of finding one proton with velocity pf=,lN2f<p{v)dvv(T. (30)
Vi and one with V2, is
By the insertion of (24), (28), (19), this becomes
iVV(viMv 2 )dvidv 2 , (27)

where <p is the Boltzmann distribution function. — \ I v2dv

In calculating the total probability of our
process, we must integrate over each pair of lire2 / 27re2 Mv2\
volume elements dvidv^ only once; if we w a n t to
integrate over v x and v 2 independently, (27)
m u s t be divided b y 2 (this corresponds to the T h e integrand has a strong maximum very
fact t h a t the total number of proton pairs is |iV 2 ). close to
• W e can transform to relative velocity v = Vi—v2 v=(±Tre2kT/hM)K (31a)
and center-of-gravity velocity V = | ( v i + v 2 ) and
integrate over the latter, then we obtain (in- Approximating it, in the usual way, by a
cluding the factor \ mentioned) Gaussian around the maximum, we find

±N2cp(v)dv (27a) p' = 16TT • 3- 5 ' 2 • gf(W)N2¥A2T2e-r (32)

with
^(v)dv=(M/47r&r)* r = 3(>ir2Me*/4h2kT)K (33)
X e x p (-Mv2/4:kT)4:irv2dv. (28)
If / is the temperature in millions of degrees,
Of the proton pairs with given velocities ViV2, we have
one in four will have opposite spins, three in
r = 33.8rV (33a)
four parallel spins. T h e latter have an anti-
symmetrical spatial wave function and therefore
§5. RESULT
do not contribute to the cross section of our
nuclear process. T h e former have a symmetrical W e insert into (32) the values of the constants
wave function which is, normalized to unit b (cf. (12b)), gf(W) from (10), A2 from Table I
density: (for ro = e2/tnc2), and assume t h a t the concentra-
tion of hydrogen (by weight) is CH and the
[(exp ( i k r r i + ^ k 2 - r 2 ) )
density (in g/cm 3 ) is p. T h e n the number of
+ (exp ( & 2 - r i + i k i . r 2 ) ) ] / v 2 . (29)
processes per gram per second is
Separating off the motion of the center of
^ = ^ / p = 167r-3- 5 / 2 -1.2-10- 5 -(6.02-10 2 3 ) 2
gravity and expanding into spherical harmonics,
• (4.37 • 10- 13 ) 3 • 8.10 -pCK2T2e-T (34)
we obtain
= 0.95-107PCH2TV-.
(exp (ikz)+exp (—ikz))/\/2
Each combination of two protons leads (§1)
= y/2sinkr/kr+>--, (29a)
ultimately to the formation of an a-particle out
which differs from (13) b y a factor of v2. This of four protons a n d two electrons, with an
gives a factor 2 in the cross section and, with the energy liberation of 4.3 - 1 0 - 5 erg. Therefore the
254 L. W. N0RDHE1M AND G. NORDHEIM

energy production is thus be considerably smaller than that at the

2 T
e = 410p£ri rV~ ergs/gram sec.
14
The density at the center of the sun is about
80, the hydrogen content £ H ^ 0 . 3 5 , therefore
PCH 2 / ^10. Table II gives the energy evolution
due to the proton-proton reaction at various
temperatures. With a hydrogen content of 35
percent, the central temperature of the sun is
about 20-10 6 degrees. At this temperature, the
energy production is about 2.2 ergs/g sec. This is
of the same order as the observed average
energy production of the sun (2.0 ergs/g sec).
Thus we come to the conclusion that the proton-
proton combination gives an energy evolution of the
right order of magnitude for the sun.
For a quantitative comparison, it must be
remembered that both the temperature and the
density of the sun decrease fairly rapidly from
the center outwards, and that the rate of re-
action decreases with both of these quantities.
The average energy production per gram will
14
Stromgren, Ergebn. d. exakt. Naturwiss. 16, (1937).
center, perhaps by a factor of 10. Since our
(35)
calculations are rather accurate, it seems that
there must be another process contributing some-
what more to the energy evolution in the sun.
This is probably the capture of protons by
carbon (reference 2).
For many problems, it is necessary to know the
temperature dependence of the energy produc-
tion e. It is convenient to express this dependence
as a power law, e^Tn. Then (cf. (35))
n = d log e/d log r = i 0 - - 2 ) . (36)
At 20 million degrees, this gives in our case
7^ = 3.5. This is large enough to make the point
source model of stars a rather good approxima-
tion. On the other hand, it is too slow a de-
pendence on temperature to explain, with tem-
peratures of the order 2-4-10 7 degrees, the very
high rates of energy production found in very
heavy stars. However, we believe that our
process is the principal source of energy in stars
lighter than the sun.

AUGUST 15, 1.9 38 PHYSICAL REVIEW VOLUME 54

On the Production of Heavy Electrons*

L. W. NORDHEIM AND G. NORDHEIM
Duke University y Durham, North Carolina
(Received June 3, 1938)

The barytron theory of nuclear forces shows a close correspondence to the theory of the
electromagnetic field. Estimates based on this analogy are given for the probabilities of the
processes leading to actual production of barytrons. A comparison with the experimental
evidence regarding the occurrence of barytrons in the cosmic radiation shows that the theo-
retical cross sections are too small to explain the properties of the hard component. The diffi-
culty is increased by the short lifetime of the barytrons which is estimated to be of the order
^ 1 0 ~ 8 sec. from radiative /3-decay. Therefore no simple picture of barytron production in terms
similar to radiation theory can be given. This failure, however, indicates only the inapplicability
of perturbation calculations, but does not constitute an actual disproof of the link between
nuclear forces and cosmic radiation.

INTRODUCTION and negative) and of a mass fx about 200 times

the electronic mass m.
A Tevidence
present there are two different lines of
for the existence of a new par-
From cosmic-ray investigations there can be
little doubt that a large percentage of the hard
ticle, the barytron, 1 of electronic charge (positive
component, which itself constitutes the greater
* A preliminary report on this paper has been given by part of all radiation from sea level downwards,
L. W. Nordheim and E. Teller at the Washington Meeting
of the American Physical Society, April 1938.
can be neither electrons nor protons2 but must
1 2
This name was proposed at the Washington Meeting of Anderson and Neddermeyer, Phys. Rev. 51, 884 (1937);
the American Physical Society, April 1938. Street and Stevenson, Phys. Rev. 51, 1005 (1937).