NEUTRON-PROTON

SCATTERINGBELOW20 MXV"

H, Pierre Noyes

Stanford Linear Accelerator Center

Stanford University
Stanford, California

*Work performed under the auspices of the U.S. Atomic Energy

Commission, in pati.while the author was at the Lawrence Radia-
tion Laboratory, Livermore
 ABSTRACT

Critical examination and analysis of existing n-p scattering

data below 20 Mev reveals that they provide quantitative information
only about the S-wave scattering lengths and effective ranges, which
are found to be

at = 5.396 + 0.011 F; a S = - 23.678 t 0.028 F;

1.726 5 0.014 F; rs = 2.51 t 0.11 k 0.043 F
rt =

where the second error quoted for r is a conservative estimate of the

S
uncertainty due to departures from the shape-independent approximation.
The correlations in error are

<&at&as>= -0.7828 6atSas ;<6at6rs>= -0.8547 6at6rs ;

<6as6rs>= 0.7029 6as6rs

An estimate of the contribution to the total cross section from scatter-

ing in higher angular momentum states, based on model calculations, p-p
phases, and,the cos 8 term in the differential cross section, allows
the deviation from the shape-independent approximation to be computed
at 14.1 and 19.665 Mev from total cross section measurements. It is
shown on theoretical grounds that this must come almost entirely from
the 'So state, and extreme limits to this variation are established.
The value found is close to zero at both energies, in accord with theo-
retical expectations, but the uncertainty is so large that it barely
excludes the extreme limits. Some qualitative evidence for or against
the existence of the long-range one pion exchange interaction in this
state could be obtained by improving the experiments below 5 Mev, hut
the uncertainty arising from the non-S wave scattering precludes any
but qualitative results. It is shown that this uncertainty cannot be
removed by improved measurement of the differential cross section be-
cause 8 independent pieces of experimental information are required.
We conclude that the energy variation of the S waves below 20 Mev can-
not be measured without recourse to experiments which separate the spin
states of the particles, such as spin-correlation, triple scattering,
polarized-beam polarized-target, etc. If some information is taken
from p-p scattering and some from theory, it might prove possible to
get away with a single such measurement in each system; this minimal
program is briefly discussed.
 I, INTRODUCTION

Although the neutron-proton interaction has been the subject of

intensive experimental and theoretical study since the discovery of the
neutron in 1932, and was correctly interpreted by Yukawa as due to the
exchange of quanta of finite mass in 1935, until very recently there
has been no basic theoretical model capable of accounting for all the
qualitative features revealed by the experimental investigations, The
discovery of two- and three-pion resonances showed immediately ly2 that
at least an import,ant part of the problem could be understood, and
connected with earlier speculations about "vector mesons." 3'4'5 It
had already been conclusively demonstrated6 that the long-range part of
the interaction in high angular momentum states is quantitatively des-
cribed by the exchange of single pions. The u) , and to a lesser extent
the PJ account for the strong short-range repulsion in the nucleon-
nucleon system, the spin-orbit interaction, and the strong short-range
attraction in the nucleon-antinucleon system. If the 'So scattering
length is fitted, single pion exchange is too weak to account for the
effective range7'8 even in the absence of a short-range repulsion, so
something must give a strong attraction in this state. Whether the
ARC phenomenon9 is due to a strong I=0 S-wave pion-pion interaction, or
is actually a resonance at a value somewhat above thresholdlo, it would
act in the nuclear-force problem like the exchange of an I=0 scalar meson
and provide this attraction; other resonance phenomena in this mass
range h5) could either strengthen or weaken this attraction, depend-
ing on their quantum numbers, but we know from the 'So parameters that
the overall effect must be attractive. We conclude 'that a minimal
description of the nucleon-nucleon interaction must contain the exchange
of the pion, of an I=0 scalar meson with a mass somewhat greater than
two pion masses, and of two (I=0 and I=l) vector mesons with about 5
pion masses. Such models have been shown by several authors 11'12'13
to give all the qualitative features found in n-p and p-p scattering
below 350 Mev in the approximation which interprets the single-particle
exchange terms as the Fourier transform of a potential, This agreement

-l-
with experiment is improved if the interaction is described by a re-
lativistic formalism (which necessitates that the p and cube treated as
Regge poles rather than as particles with a discrete mass and angular
momentum); then only 9 parameters, five of which can already be roughly
estimated from other phenomena, are needed to make this agreement near-
ly quantitative over the entire energy range.
If, as this author believes, this signal success is due to the fact
that the most important physical phenomena responsible for the two-
nucleon interaction have finally been isolated and partially understood,
and not just a misleading accident, future work on the two-nucleon pro-
blem will differ radically from the generally frustrating confusion
which has characterized this field in the past.l' For one thing, there
will now be considerably more incentive for including the mesonic
degrees of freedom in the study of nuclear matter, and some hope of
success. As noted by Teller16 some time ago, the fact that the spin-
flip iso-spin-flip one pion exchange is forbidden to first order by the
Pauli principle in nuclear matter, implies that the dominant interactions
will be due to the scalar and vector meson exchanges we discussed above;
consequently Duerr's I7 interpretation of the Teller-Johnson model la
has finally been connected up with elementary-particle physics in a
qualitative way, and pursuit of the quantitative connections might
prove revealing. A second area where work will now go forward is the
determination of the parameters of the resonances from nucleon-nucleon
scattering data, and calculation, or at least estimation, of the non-
resonant background.lg Unfortuna tely it appears unlikely at present
that these parameters can be computed from pion-nucleon or'pion-pion scattering
to the accuracy required for a quantitative fit to the nucleon-nucleon
data; consequently this work will provide a consistency check rather
than a quantitative test of the theory. To make quantitative tests of
the model it will be necessary to tie down the short range parts of
the interaction (coming from kaon, hyperon, n-pion, .,. exchanges) by
phenomenological parameters determined at low energy, and test the
theory by comparing the energy variation of the scattering amplitudes
predicted by the longer range parts of the interaction with experiment.

-2-
Unfortunately this will require more parameters than the S-wave scatter-
ing lengths. Breit and Hull 2o have shown that centrifugal shielding
of the P-waves is incomplete, and consequently that the 'P and iP
0JlJ2 1
phase shifts cannot be accurately computed from a knowledge of the
long-range part of the interaction at any energy; we therefore will
require four P-wave scattering lengths to be determined from experi-
ment. Because of the strong tensor force, the 'S - 3D coupling para-
meter 21 e1 and 'D phase shift 6 will also beiinflienced by the
1 2’1
short-range part of the interaction in the 3S state, and we will
1
need two empirical constants for these states. We conclude that in
order to utilize scattering data at high energy for quantitative
tests of the theory of the n-p interaction it is first necessary to
determine 8 constants from low energy scattering experiments. The
remainder of this paper is devoted to the study of what constants
can be determined from existing experiments, and what additional
experiments might be required for this purpose.

II. THE SHAPE-INDEPENDENTAPPROXIMATION

Since n-p scattering below 20 Mev is dominated by the two S-waves,

our first concern will be to isolate these two amplitudes and character-
ize them as accurately as possible. The first step is to assume that
they are given by the shape-independent approximation

T = k ctn 6 = -l/at + 3 rtk2

0’1

S E k ctn 6 0 = -l/a s + 3 rsk2

- 3 r
so called because any interaction containing two adjustable parameters
reasonably sensitive to the overall strength and range (or dependence
on momentum transfer) can be fitted to this energy variation at suffi-
ciently low energy. The scattering lengths at and as can be determined
directly by the measurement of the coherent neutron-hydrogen scattering
length, anII9 and the total n-p cross section, 0 measured at energies
0’
just above the point where molecular effects become significant sources
of uncertainty, since

+3at)=-*a
"nH = $(a,
(2)
u = r( (a: + 3az) = fi C
0

In connection with the analysis of their precision n-p total cross

section measurements at 0.4926 and 3.205 Mev, Engelke, Benenson,
Melkonian, and Lebowitz " have made a critical survey of the existing
measurements of u. and conclude that
the best value to adopt is that
given by Melkonian2' of cJo= 2036 +, 5 F'?. After discussion with
Engelke,24 it appears that the best value of anR is the weighted-mean
given by Richard Wilson 25 as an, = 3.744 C 0.010 F. Hence from
Eqn. 2 we have that

at = $ b - a) = 5.397 f 0.011 F

a
s = -+ (3s +a) = - 23.679 5 0.028 F
with (3)
s = [ (4 c - a”) /3+

The correlation in error is given by

< Sat 6as > 1 L

SC2 - -&-(3s + a>( s - a)6ae /12C

<sa2
t >h<6a..
s>*=[ (SF$-$(5- a)%?)pJf [&-2++( 3s + .)%?‘)/36s’]+
= - ( ) . ’ (7 : ( .I j

: )t -
 In order to determine the effective ranges rs and rt we must
make use of experimental information at finite values of k", where fik is
the momentum of either particle in the c.m. system. For neutrons in-
cident on a stationary proton target, this is given by26

2K
s2k2 = lab Mz %
M,(M, + Mpj2 (4)
k2 = 0.0120484 Klab F-2

where Klab is the energy of the incident neutron in Mev. Since the
*.
deuteron corresponds to a pole 27 in the scattering amplitude e lbO,l
sin LZj = l/(T - ik) at K2k02 = - 2MnMped/(Mn + MP) where sd is the
‘J1
binding energy, we can evaluate the triplet effective range as

=2(1- l/atko l/k

't 0
(5)
= 1.727 + 0.014 F

where we have used the latest measurement of the binding energy of the
deuteron by Knowles 28 of ed = 2224.52 + 0.20 kev. The accuracy of
this measurement is so high that the uncertainty in rt arises solely
from the uncertainty in at, and it is easy to show that the uncertainty
in T due to ed is less than 0.13% of the uncertainty due to a
at any
t -
energy; we can therefore take k as exactly known in what follows.
0
In order to obtain a reliable value of rs from total cross section
data a-$ low energy, it is crucial to select from the mass of existing
data those experiments which are most likely to be free from system-
atic error. This thankless task has been performed for me by Hafner ";
the experiments, selected on the basis that they are known to be free
of systematic error due both to in scattering and to neutrons degraded in energy
by other processes,are given in Table I, to which havebeen added the
new measurements of Engelke, et.al. 22 Hafner also provided a larger
selection containing about 20 more measurements, and the analysis pre-
sented below has also been carried through for these using various
selections; since the results are insensitive to the addition of these
measurements to those given in Table I, and the errors are not signifi-
cantly improved, we will give results only for the smaller selection.
Richard Wilson 25 concurs with the selections made by Hafner and their
evaluation.
The values of at9 as, and rs are determined by adding the two
experiments already discussed and minimizing

ps + jat>- anlI
I bazH
+ -2

3x
+
T2 + k2

The error matrix is obtained by calculating the inverse to 3 & , where

j
Xi' x. run over the three parameters and the second derivative is com-
J
puted at the minimum. We find that the values of as, at and their errors
are essentially unchanged from those given above (final values are given
in the Abstract), and that

r = 2.51 + 0.11 F (7)

S
with error correlations

< 6rs Eat > = - 0.8547 < 6rE > + < sat >+
(8)
1
< 6rs 5as > = 0.7029 < 6rz > 3 < 8az >2

Before turning to a discussion of the uncertainties in these values

due to departures from the shape-independent approximation, we wish to
discuss this value for rs. To begin with, we note that it is signifi-
cantly different from the p-p effective range r = 2.78~ given by
PP
Heller 31. This deviation from charge independence is greater than
one would expect from the 3-$$ n' - 51' mass difference, but since we have
seen above that the "scalar meson" is more important than single pion
exchange in determining the singlet effective range, it will be impossible
to discover whether this is a real failure of the charge-independence
hypothesis until the structure of this object is sufficiently well
understood to allow a calculation of the electromagnetic corrections
to its effective mass and coupling constant in the n-p and p-p systems.
It is perhaps worth noting that if this charge-dependent value of
r w = 2.3 F is accepted, the discrepancy between the computed and
S
observed value of the n-p t'nermal capture cross section nearly disappears.32
We note further that the two new measurements at 0.4926 and 3.203 Mev
by themselves would gife
rs= 2.43 t 0.11 F 33 while the six earlier
measurements gave 2.64 + 0.12 F. This looks a little large for a
statistical fluctuation and suggests that additional precise measurements
of n-p total cross sections in this energy range would be of value. As
we will see in the next section, however, there is no point in pushing
the precision of such measurements beyond the point already achieved
by Engelke, et. al. unless the precision of a nH and the very low energy
cross section is also improved.

III. DEVIATIONS FROMTHE SHAPE-INDEPENDENTAPPROXIMATION

Until we can set a-priori limits to the deviations from the shape-
independent approximation as a function of energy we can neither assess
the reliability of the parameters determined in the last section nor
determine the requisite accuracy for experiments at higher energy which
would give significantly new information about the energy dependence
of the S waves. These deviations will come from two sources. In the
first place, we can anticipate k4 and higher terms in the exact express-
ions for S and T, and must estimate the magnitude of their coefficients.
Since these terms will cause the total cross section to deviate from
the approximation to order k4, and e1 and the P phase shifts will con-
tribute terms of the same order, we must also be able to estimate the
contribution to the total cross section from higher angular momentum
states. We will start with this second problem.
As discussed in the Introduction, six phase parameters other than
the S phase shifts cannot be predicted from one-pion exchange (OPE) at

- 7 -
any energy. Sin02 existing data below 20 Mev consists only of differen-
tial and total cross sections, we cannot at present take these from
experiment. For the 'S -3D 1 state we believe that the models of
Glendenning and Kramer 34 , which consfst of an 0P.E tail and an inner
phencmenological part fitted to the deuteron,and which are in rough
agreement with n-p scattering analyses at high energy, should give a
reasonable estimate. Dr. Glendenning 35 has kindly supplied me with
phase shifts for these models at 1, 5, 10, and 14.4 Mev, and I find
that the contribution to the total cross section from E' and 6 differs
very little between the various models. For the 'P phases, we'iisume
charge independence and take them from the energy-dependent phase
shift analyses of Stapp, et.al. 36 The contribution from these triplet
phases is given in Table II. While the phase shift values themselves
are not particularly reliable, we see that the spread between the cross
section contributions is so small that we can perhaps believe .the order of
magnitude of the total c.ross section prediction.
1
For the Plstate, we note that existing models and theories agree
that whatever interaction is present in addition to OPE it is also
predominantly repulsive. Since in effect this additional interaction
simply strengthens the centrifugal barrier we can expect much smaller
deviations from the OPE value than if either it or the short-range
interaction were attractive. It is possible to make a rough check on
this theoretical prediction in the following way. The strong tensor
force in the triplet state leads to much more isotropic scattering
than would be expected from impact parameter arguments 37. Consequently,
to a first approximation the angular variation
of the differential
1 1
cross section is dominated by the cos 8 term arising from So- Pl inter-
ference. As the 3S1 phase shift is accurately given by the shape in-
dependent approximation in this energy range (cf. below) we can obtain
6, from the total cross section and hence evaluate E1 from the cos 8
term in the differential cross section measurements at 14.1 38'3g and
17.9 4o Mev. We have actually carried through this analysis *I, finding
that in fact the phase shifts used for Table II do give the expected
approximate isotropy in the triplet scattering, and obtained values of

-8-
E1 at these two energies which are consistent with OPE. Unfortunately
the error is about 35’$ in sin 83. , which precluded a reliable extrapola-
tion to 1().66:j Mev. WC thercforc feel it more reliable to use the OPE
value, but assign an experimental uncertainty of >O$, to sin '% which
we believe to be conservative. Since we need in addition value: only
below 5 Mev, we made a rough extrapolation by assuming that the energy
variation of the phases was the same as for OPE and obtained the values
given in Table I. Since this estimate is only 20% of the experimental
error for the highest energy in the Table, we believe we have success-
fully eliminated this source of uncertainty from the analysis given
in the last section.
Most discussions of the departure of the S-waves from the shape-
independent approximation make use of a flshape parameter" P defined
by adding a term -P?k4 to the expressions given in Eqn. 1. This is
inadequate for our purposes since the interaction due to OP?3gives a
branch point in S and T at k2= - (m,c/wl)' corresponding to a "laboratory"
energy of -10 Mev; consequently the expansion in powers of k2 about k2= 0
diverges beyond 10 Mev and is quantitatively unreliable at much lower
energies. As was shown by Noyes and Wang', it is possible to take
account of OPE exactly and extend the a - priori radius of convergence
to 40 Mev at the cost of solving a non-singular integral equation of
the Fredholm type. In the approximation which replaces the multi-particle
exchange branch cuts by a single pole whose position and residue are
adjusted to fit the observed scattering length and effective range,
the solution to this integral equation is very accurately represented
(to better than 7% up to 40 Mev) by a simple expression derived indepen-
dently by Cini, Fubini, and Stanghellini 42 from fixed angle dispersion
relations. This is equivalent to replacing the OPE branch cut by a
single pole of known residue at k2 = -$(m,c/fi)2 = -$-rom2 and hence to
the expression

c (kr0J4
kro ctn 6 = a + b (kro)2 + (9)
1 + d(kro)2

-9-
 2- f2~ (c$\/?. + &a - b)
d=
1 - fZM ($ 4% + a)

c= - ( 1 - $d)(2\/ - 2 b + lka)

where M is the ratio of the nucleon to the pion mass and f2(= 0.08) the
pion-nucleon coupling constant. Since we are given the triplet scatter-
ing length and deuteron binding energy rather than the triplet effective
range, it is convenient in the triplet case to introduce the pole at
k2 = - qo2/rt explicitly, which can be done by taking b = (l/s,)x (1 + a/qo)
+ &/(l - dqz). Taking for m7( a third of the neutral pion
7
mass ard two-thirds of the charged pion mass, and f2M = 14.4 Mn) =
1
0.529% we find for the triplet case

r = 1.4292 F a = -0.26481 b = 0.60350 c = - 0.04389 d = 1.8439 (10)

0

We therefore predict for the triplet shape parameter the small value
Pt = - c/8b3 = 0.025, in addition to an important damping of this term
by the denominator 1 + d(kro)2 in the 10 - 20 Mev range.
In order to strengthen our belief in the small size of this shape-
dependent term, we have compared the phase shifts computed by Glendenning 35
with the shape-independent approximation, and in all cases find the
deviation to lie within the theoretical estimate. In making this com-
parison it was important to note that the values of the scattering
length differ for the different models and to use the appropriate
value in each case. Since these models are in reasonable agreement with
experiment at high energy, we believe that the magnitude (although not
the sign) of the triplet shape dependence is conservatively estimated
by Eqns. (9) and (10).
For the lSo state, there is no longer a pole at k2 = - ko2 but instead
a pole on the second Riemann sheet corresponding to a virtual rather
than actual bound state. We therefore take b = 3 r,/r, and, for rs= 2.544
determined by a least squares adjustment (cf. below) find that
a = 0.060358 b = 0.89013 c = -0.28312 d = 1.5629 01)

- LO -
predicting a singlet shape parameter Ps = 0.050. This prediction is
considerably less reliable than for the triplet state because, as already
noted, the singlet state is much more strongly influenced by the scalar
and vector mesons than by OPE, the repulsion due to the vector mesons
causing the singlet phase shift to change sign at 250 Mev, a behavior
not predicted by the CFS approximation. If two additional empirical
constants are added to t‘ne integral equation and fitted to high energy
phase shifts we find that close to k2 = 0, the estimate of Ps is still
approximately correct, but that k ctn Go crosses the shape-independent
approximation around 20 Mev. This behavior is sketched in Figure 1.
Lacking the requisite experimental information, we again turn to
model calculations for confirmation of this prediction and find that
models which have either a hard core or boundary condition to fit the
250 Mev singularity, the OPE tail, and something else to fit the values
of a and rs do indeed behave in this way 43'44 but that the cross-over
S
point is sensitive to the details of the model and may occur anywhere
between 10 and 40 Mev. We believe that this shows that the CFS curve
gives a conservative estimate of the amount by which the actual curve
is likely to fall below the shape-independent approximation at low
energy.
In order to estimate the amount by which we can expect the curve
to lie above the shape independent approximation in the extreme case,
we make use of the two-parameter model which has no
- potential (OPE) tail,
but consists of an energy-independent boundary condition on the wave
function at finite radius 45'46

k ctn (6 + kF> = A 02)

Making a least squares adjustment we find that r = 1.1701 F and

A = 0.040245 F-i. This gives a curve with a singularity at about 135
Mev as indicated in Figure 1, and again gives what we believe to be a
conservative estimate of the deviation. This is confirmed by adding
an energy dependent term to Eqn, 12 to move the singularity out to
250 Mev, which results in a curve which lies everywhere between the
BC curve and the shape-independent approximation and always above the
- 11 -
latter. We can therefore obtain an estimate of the shape dependence of
the value of r determined in the last section by adjusting either the
S
CFS or the BC models to the same data. Since we have added no more
free parameters, the error analysis is unaltered, and we find

CFS
r = 2.544 f 0.11 F rBC = 2.4% + 0.11 F (13)
S S

We conclude that in addition to the experimental error of 0.11 F we

should add an additional uncertainty due to shape dependence which is
at most + 0.043 F.
The results of this discussion are summarized in Figure 2 where
we plot separately the various sources of uncertainty in the determina-
tion of rs. These are (a) uncertainty due to the scattering lengths,
(b) uncertainty due to the triplet shape dependence, (c) uncertainty
due to the contribution from other angular momentum states, (d) un-
certainty due to the singlet shape dependence, and (e) for reference,
the uncertainty due to a 1 mb error in the measurement of the total
cross section. Several conclusions follow immediately from this curve.
As already discussed by Engelke, et. al. 22, their low energy point
is located at the optimum energy for the determination of rs, and their
experimental uncertainty has been reduced to the point where it is
equal to the experimental uncertainty arising from other causes (mainly
the scattering lengths). Any improvement will therefore require in
addition remeasurement of the scattering lengths to higher precision.
As we go to higher energy we see that the 15 - 20 Mev region is the
most favorable from the point of view of the error due to the scatter-
ing lengths; this is due to the fact that
k ctn 60,-1 goes through zero
at about 17.8 Mev so that the sensitivity to a is'very small, and that
t
this is already at a high enough energy so that the sensitivity to as
is nearly negligible. We further see that the 2 - 23mb error in our
knowledge of the scattering in other angular momentum states is serious
and precludes much more than a qualitative distinction between the two
extreme singlet curves. From our discussion given above it is also
clear that this uncertainty can only be slightly reduced by more accurate

- 12 -
measurements of the cos 8 term in the differential cross section, and
consequently that measurements which distinguish the polarization states
of the neutron and proton and lead to a phase shift analysis are required
to make a quantitative determination of the variation of the S waves
away from the shape independent approximation. This is the most im-
portant result obtained in this paper. We will discuss briefly what
is required in the next section.
Finally we ask what the cross section at 14.1 Mev of 689 -+ 5 mb
measured by Poss, Salant, Snow and Yuan 38 and at 19.665 Mev of 494.2 + 2.5 mb
measured by Day, Mills, Perry and Scherb 3g can tell us about the de-
viation of the S-waves from the shape-independent approximation. As
already discussed, this cannot give us the shape parameter defined at
k2 = 0, so instead we test the result against the three extreme two-
parameter models (BC,SI, CFS) by computing the shape function 47 at
this energy from the deviation of the total cross section from the
shape-independent approximation. Clearly, the formula is

(S2 + k2)2
P(k2) = %ot 04)
2rrSk4r3S

For consistency we must compute aSI separately for each model due to
the differences in rs, and we must be careful to include the correlations
in error in calculating the uncertainty. We also include the uncertainty
due to the triplet shape parameter. The predictions and errors are
collected in Table III. The corresponding predictions
and observations
of the shape function at these energies are given in Table IV. We see that
there is no significant deviation from the shape independent approximation.
While the value of P(k2) close to zero is in accord with our theo-
retical expectations, we see that the errors are still too large to
give any significant discrimination between the extreme models.
(Cf. Figure 2). We note also that if the value of rs were 2.43 F, these
results would strongly favor the BC model, while if it were 2.64 F we
would say that the BC model was pretty conclusively excluded, emphasizing
- 13 -
again the necessity for improving our confidence in r S by new measure-
ments at low energy. Finally we reiterate that improved total and
differential cross section measurements below 20 Mev can at best decide
between the extreme models and can never give the detailed energy
1
variation of Ps(k2) needed to test theories of the So scattering in
this energy range.

IV. CONCLUSIONS

Evaluation of all the important uncertainties reveals that our

knowledge of the n-p S-wave scattering lengths and effective ranges
could be improved by at most a factor of three over the values given
by this analysis, if the accuracy of total cross section measurements
below 'j Mev, of the very low energy total cross section, and of the
coherent neutron-hydrogen scattering lengths were improved by that
amount. Such additional measurements would also be desirable because
the spread in existing measurements is somewhat larger than is to be
expected on purely statistical grounds and makes it questionable
whether the mean value of r obtained by this analysis can be trusted
S
to the quoted statistical accuracy. If the low energy analysis is
accepted, the total cross section measurements at 14.1 and lg.663 Mev
.
give a very small singlet shape effect at these energies, in accord
with theoretical expectations; this confirmation would become more
convincing if the above improvement in the lower energy measurements
were achieved. However, our current lack of knowledge of the scatter-
ing in other angular momentum states at these energies is,comparable
to the experimental uncertainty in the total cross sections, and we
have shown that this uncertainty cannot be removed by improved measure-
ment of the differential cross section. We conclude that quantitative
information about the departure of the S waves from the shape indepen-
dent approximation can be achieved only by performing enough new types
of experiments to lead to a unique phase shift analysis.
The n-p cross section at low energy depends on eight Fhase ?ara-
meters which cannot be evaluated from one pion exchange, so in principle

- 14 -
eight independent pieces of experimental information are needed at each
energy where these phases are to be determined. If we are willing to
3 o 1 2 phases could be taken from p-p
assume charge independence, the P
scattering. Since in that system t;e'Coulomb interference terms in
the differential cross section give three independent pieces of infor-
48
mation , one needs in addition one experiment such as Cnn (90”) to
1
obtain So and these three P phases, as has been discussed by Iwadare *'.
Actually the P phases are still given only up to a four-fold trigonometric
ambiguity, but since the analyses are unique at higher energy, and in
agreement with the theoretical prediction 50, an additional experiment
such as D or A is needed to resolve this ambiguity only to the extent
that one distrusts the extrapolation or the theoretical argument. Charge-
dependent effects are still big enough at 20 Mev so that we cannot
reliably use the 'So phase determined from p-p scattering to assist
the n-p analysis for experiments of the precision contemplated here;
we therefore still need five pieces of information from n-p experiments.
Two of these can certainly be provided by the total cross section and
the cos 8 term in the differential cross section if improved measure-
ments of the latter are made. The polarization has recently been
measured to high precision at 23.1 Mev 'i, so we can count on this for
at least one more piece of information. If one accepts the theoretical
argument of Wong 52, s1 can be calculated to the requisite accuracy
from a knowledge of 3Sl and the OPF,interaction, so a minimal program
would require only one of the difficult experiments (spin correlation,
triple scattering, polarized beam-polarized target, etc.), but de-
tailed examination of the requisite accuracy will not be attempted
here. Preliminary calculations 53 indicate that the spin-correlation
and to a lesser extent the depolarization experiments are more sensitive
to small variations in the phase shifts than the R or A parameters.
Since phase shifts computed from the Hamada-Johnston model 54 are in
excellent agreement with the polarization measurement 51, they should
provide a reliable starting point for the optimization of the experi-
mental design. Ultimately one hopes that a sufficient variety of ex-
periments will be performed to lead to unique phase parameters from

- 15 -
n-p experiments alone, and hence to the eight low-energy empirical con-
stants discussed in the Introduction.

ACKNOWLEDGEMENTS

The author wishes to take this opportunity to express his gratitude

for the unfailing support and encouragement given him for this and other
research on the nucleon-nucleon problem by S. Fernbach of the Lawrence
Radiation Laboratory, Liver-more. This paper would have been impossible
to carry through without the critical evaluation of the n-p data pro-
vided by E.M.Hafner. Advise and criticism on experimental points and
analysis from R.Wilson and C.E.Engelke are gratefully acknowledged.
Assistance with the extensive preliminary numerical work was provided
by E.Degraw and the Computation staff at LRL.

- 16 -
 REFERENCES
1. H.P.Noyes, Proc. Rutherford Jubilee International Conf., p. 749,
Heywood and Co., Ltd. (1962).
2. G.Breit, Proc. Rutherford Jubilee International Conf., p. 756,
Hey-wood and Co., Ltd. (1962).
3* G.Breit, Phys. Rev. 2, 248 (1936); S.Share and G.Breit, Phys. Rev,
52, 546 (1937); G.Breit, Phys. Rev. 53, 153 (1938); G.Breit and
J.R.Stehn,Phys. Rev. L53, 459 (1938); G.Breit, Proc. Natl. Acad.
Sci. U.S., 5, 746 (1960), G.Breit, Phy. Rev. 9, 287 (1960).
4. Y.Nambu, Phys. Rev. 106, 1366 (1957).
5. J.J.Sakurai, Phys. Rev. 2, 1784 (1960); Ann. Phys. 11, 1 (1960).
6. For a review of this evidence and references to earlier work cf.
M.J.Moravcsik and H.P.Noyes, Ann. Rev. Nut. Sci. 11, 95 (1961).
7* J.Iwadare, S.Otsuki, R.Tamagaki, and W.Watari, Pro@;. Theor. Phys.
I-J, 86 (1956); Prog. Theor. Phys. I&, 472 (1956).
8. H.P. Noyes and D.Y.Wong, Phys. Rev. Letters 2, 191 (1959):
9. A.Abashian, N.E. Booth, and K.M. Crowe, Phys. Rev. Letters 1, 35 (1961).
10. B.Richter, Phy. Rev. Letters 2, 217 (1962). R.Del Fabbro, M.De Pretis,
G.Marini, A.Odian, G.Stoppini, L.Tau, and R.Visentin, unpublished
report at the 1962 High Energy Physics Conference at CERN.
11. R.S.McKean, Jr., Phys. Rev. 3, 1399 (1962).
12. D.B.Lichtenberg, J.S.Kovacs, and H.McManus, Bull. Am. Phys. Sot. 1,
55 (1962).
13. R.A.Bryan, C.R.Dismukes, and W.Ramsay, UCLA preprint.
14. A.Scotti. and D.Y.Wong, submitted to Phys. Rev. Letters and private
communication.
15. To quote M.L.Goldberger "There are few problems in modern theoretical
physics which have attracted more attention than that
of trying to
determine the fundamental interaction between two nucleons. It is
also true that scarcely
ever has the world of physics owed so little
to so many...... In general, in surveying the field, one is oppress-
ed by the unbelievable confusion and conflict that exists. It is
hard to believe that many of the authors are talking about the
same problem, or in fact that they know what the problem is.)'
Proc. Midwest Conf. Theor. Phys., 50-63 (Purdue Univ., Lafayette
Ind., April, 1960).

- 17 -
16. E.Teller, private communication.
179 H.P.Duerr, Phys. Rev. 103, 469 (1956).
18. M.H.Johnson and E.Teller, Phys. Rev. 98, 783 (1955).
19* The non-resonant effects are certainly required for a consistent
treatment of the problem and the calculations of D.Amati, E.Leader,
and B.Vitale, Nuovo Cimento (in press), and of W.N.Cottingham and
R.Vinh Mau, Phys. Rev. (in press) indicate that this contribution
to the interaction could be comparable in magnitude to that coming
from the resonances.
20. G.Breit and M.H.Hull, Jr. Nuclear Phys. 3, 216 (1960).
21. Throughout we use the nuclear-bar parameters btJ and eJ as defined
by H.P.Stapp, T.Ypsilantis, and N.Metropolis, PAYS. Rev. 105, 311
(1957).
22. C.E.Engelke, R.E.Benehson, E.Melkonian, and J.M.Lebowitz, Phys.
Rev. (in press).
23. E. Melkonian, phys. Rev. -76, 1744 (1949); we follow Engelke et.al.
in using the standard deviation rather than twice that figure as
quoted by the author.
24. C.E.Engelke, private communication. I am indebted to Dr. Engelke
for informing me of his results prior to publication and for de-
tailed discussion of our somewhat different analyses.
25. R.Wilson in a forthcoming book on nucleon-nucleon scattering
experiments to be published by Interscience; I am indebted to
Prof. Wilson for sending portions of the manuscript and for
discussion of several points.
26. We use the values given by W.H.Barkas and A.H.Rosenfeld, UCRL -8030
(rev.) of Mpc2 = 938.213 Mev, M,"' = 939.507 Mev, ?ic = 197.32 Mev F;
note that using the average mass is not quite accurate enough as
it would change rs by about 0.02 F. However the exact relativistic
expression, which is obtained by multiplying Eqn. 4 on the right
by (1 +K/2M,c')/ p 2MpK/(Mn+Mp)2c2], since it is proportional
to the n-p mass difference (if K is the difference between the
total neutron energy and its rest energy), differs by only about
one part in lo6 from Eqn.4 in this energy range.

- 18 -
27. M.L.Goldberger, Y.Nambu, and R.Oehme, Ann. Phys. 2, 226 (1957).
28. J.W.Knowles, Can. J. Phys. 40, 257 (1962).
29. R.Howerton, UCRL 5226.
30. E.M,Hafner, private communication.
31. L.Heller, Phys. Rev. 120, 627 (1960). Recent work by the author (H.P.N.,
unpublished) indicates a somewhat lower value, but definitely
greater than 2.7 F.
32. N.Austern and E.Rost, Phy. Rev. 117, 1506 (lg60), Eqn. 16.
33. The two values are 2.448 and 2.374 F at 0.4926 and 3.205 Mev res-
pectively; the first differs from the value of 2!,46 F quoted in
reference 21 because we have used a different value for a *
nH'
comparison shows that the difference is due solely to this dif-
ference in input.
34. N.Glendenning and G.Kramer, Phys. Rev. 126, 2159 (1962).
35. N.Glendenning, private communication.
36. H.P.Stapp, H.P.Noyes, and M.J.MoravcsLk, in preparation.
37. R.S.Christian and H.P.Noyes, Phys. Rev. 79, 85 (1950).
38. H.L.Poss, E.D.Salant, G.A.Snow, and L.C.L.Yuan, Phys. Rev. 87, 11
(1952).
39. R.B.Day, R.L.Mills, J.E.Perry, Jr. and F.Scherb, Phys. Rev. 12,
209 (1959).
40. A.Galonsky and J.P.Judish, Phys. Rev. 100, 121 (1955); I am in-
debted to Dr.Galonsky for sending me the numerical results of
this experiment for analysis.
41. We were assisted in the numerical work by D.Quinn.
42. M.Cini, S.Fubini, and A.Stanghellini, Phys. Rev. l&, 1633 (1959).
43. P.Signell and R.Yoder, Phys. Rev. 12;1, 1897 (1961).
44. J.K.Perring and R.N.J.Phillips, Nuclear Phys. 23, 153 (1961).
45. G.Breit and W.G.Bouricius, Phys. Rev. 75, 1029 (1949).
46. H.FeshbachandE.Lomon, Phy. Rev. 102, 891 (1956).
47. H.P.Noyes, Bull. Am. Phys. Sot. 1, 504 (1962) and above. The
numbers quoted there have changed somewhat due to changes in
input data. The positive value of P given is obtained primarily
1
because the Pl phase shift at 14.1 Mev was taken from the

- 19 -
 differential cross section analysis rather than OPE; we now
believe the latter to be more reliable for the reasons given
in the text.
48. E.Clementel, C.Villi, and L.Jess, Nuovo Cimento 2, 907 (1957).
49. J.Iwadare, Proc. Phys. Sot. 78, 185 (1961).
50. Riazuddin, Phys. Rev. 121, 1509 (1961).
51. R.B.Perkins and J.E.Simmons, Los Alsmos Preprint.
52. D.Y.Wong, Phys. Rev. Letters 1, 406 (1959).
53- A.DuBow and D. Halda, private communication.
54. T.Hamada and I.D.Johnston, Nuclear Physics 2, 382 (1962).

- 20 -
 TABLE I.

Neutron-proton total cross section measurements below 5 Mev and the

estimated contribution to the total cross section coming from angular
momentum states other than 'So and 3S1.

Point Reference Energy (Mev) Cross section &f 0 CT;;;ibution

(mb)
1. o. 4926 6202 t ii 0.03
2. 1.005 4228 18 0.17
3. 1.078 4060 30 0.22
4. 1.315 3675 20 0.29
5. 1.578 3330 20 0.39
6. 2.540 2525 9 0.69
7. 3.205 2206 7 0.89
8. 4.749 1690 6 1.20

aEngelke, et.al., Reference 22.

b C.E.Falk, G.Snow, and T.Coor, Phys. RevpQ,
E.M.Hafner,,W.F.Hornyak,
204 (1953).
'E.E.Lampi, G.Freier, and J.H. Williams, Phys. Rev.76, 188 (1949).
d
C.L. Storrs and D,H.Frisch, Phys. Rev. D, 1252 (1.954).
e
R.E.Fields, R.L.Becker, and R.K.Adair, Phys. Rev. -94, 39 (1.954).

- 21 -
 I

TABLE II

Estimate of @ 0 contribution to the total

n-p cross section between
14 and 20 Mev. 3s 1- 3D state taken from Glendenning, 35 3p
l1 1 0,1,2 from
Stapp, et.al. " and Pl from OPE; the error in Pl is estimated from
differential cross section measurements at 14.1 and 17.9 Mev.

1
Energy(Mev) T~~;le~Dcon~;ibuti~n
pl
1- 1' 0,1,2

14.1 Mev 1.27 + 0.17mb 0.76 mb 3.85 + 1.92 5.83 t 2.10 mb

17.9 2.47 0.10 1.10 4.39 2.19 7.95 2.29
1-g.665 2.81 0.15 1.25 4.56 2.28 8.63 2.43

- 22 -
 TABLE III.

Contributions to the n-p total cross section at 14.1 and lg.665 Mev.

Laboratory rs from Shape-independent Triplet shape Scattering

Energy prediction uncertainty for $+ 0

14.1 Mev BC 684.832 ?I 2.327 + 0.598 5.834 5 2.096 mb

SI 683.460
CJTS 682.696

19.665 Mev BC 487.423 + 2.308 + 0.260 8.627 + 2.43 mb

SI 486.288
CFS 485.657

- 23 -
 , TABL;E IV

1
Shape function for So n-p scattering at 14.1 and lg.665 Mev.

14.1 Mev. lg.665 Mev.

Model Predicted Observed Predicted Observed

BC -0.041 -0.012 -0.042 -0.012 + 0.027

SI 0 -0.002 + 0.045 0 -0.005
CFS 0.026 0.004 0.024 -0.001

- 24 -
 FIGURE CAPTIONS

Figure 1. Predictions for the effective range function defined by

rs(k2) = 2 ( k ctn ho + l/as)/k2. The expected behavior for models
which have the long range OPE interaction and sufficiant parameters
tofitas, r s (0), and the zero in so at 250 Mev is giri-en by the
curve MCFS. The three other curves are CFS (Cini, Pubini, and
Stanghellini 42), SI ( s hape-independent approximation), and BC
(boundary condition 4sy46).

Figure 2. Contributions to the uncertainty in the value of rs comput-

ed from total cross section measurements at a single energy: + 6a due
to uncertainty in the scattering lengths,
rt 6 0 due to the
Go
scattering in other angular momentum states, -+ 6Pt due to the triplet
S shape dependence as estimated by the CFS formula, 2 6 (r due to an
experimental uncertainty of one or 10 mb in the total cross section.
The curves BC and CFS give the maximum theoretically allowed devia-
tion from a constant value (cf. text). The arrow heads on the two
experimental points show the uncertainty arising from the exper-
imental error in the total cross section. The point at 19.665 Mev is
repeated to the right of the graph to show the effect of the data
selected below 5 Mev to determine rs(0), and of the extreme varia-
tion arising from the shape dependence assumed in that analysis.

- 25 -
 I
I
I
I
I
I
I
I
I
I
I
I

-I .
I I II I I
100 200
LABORATORY ENERGY IN MEV

FIG. 1
- 26 -
 rs (0) FROM DATA AT :

0.4 .4925 MLV I.001 MCV

5.205 I.078
I.313
COMBINED
I.578
2.S4ii
0.3 4.149
T
0.2

I
z
- 0
r:
I r

- 0. I
>
b
‘Z

5
K
w -0.2
x
3

SHAPE ASSUMED BELOW 5 MLV

- 0.4 51 SI SC CFS
5 10 IS
LABORATOQY ENERGY IN MEv

FIG. 2