1.

31 ANALYTICITY AND CROSSING 13

must have real positive energy pro) for each particle, and real three-
momentum pz. fn the equal-mass case this gives

s>-4rn2, ,<0, zc (0. (1.3.e)

This result can be obtained by expressing s,t,'tt interms of the momen-

tum q and the scattering angle d in the centre-of-mass system, which
gives
s:A(mz*gr\, )
r, - (1.3.10)
-2q'(L- "o, d), [
q,: -2q'(L+cosd).J
When the masses are not equal the conditions are not quite so simple;
they are derived by Kibble (1960) (see also S4.3).
So far we have varied only s in discussing analytic continuation, but,
in general both s and t can be regarded as complex variables in the
amplitude -F (s, f). Then we can consider analytic continuation from the
physical region (I.3.9) to the region

u)4m2, s(0, ,<0. (r.3.rr)

It is assumed that the resultant function ,t', evaluated in a suitable
limit on to the region (l.3.tl), is the physical scattering amplitude for
the Process ar*A"-Ar+An, (r.B.r2)

where An denotes the anti-particle of An. Eot this process the energy
in the centre-of-mass frame for the initial (or final) state is just {tr.
It is further assumed that by analytic continuation to the region
t ) 4m2, xe ( 0, I ( 0, (1.3.13)

the function .F, evaluated in a suitable limit, gives the physical

scattering amplitude for the process

Ar* An- Ar+ A, (r.3.14)

for which the energy in the centre-of-mass frame i*

These important properties are called the 'crossing' ^ft. properties.
They state that the same analytic function can be used to describe the
three different physical processes (1.3.8), (1.3.12), (1.3.14) by making
anappropriate choice of physical values for the variables s and t (or u).
These physical processes are often called different 'channels', and one
refers to them as the s-channel, the f-channel and ttre u-channel when
14 ANAr,yrrcrry AND cRossrNc [1.3
s, f and ?r, respectively, are the energy variables. Remembering the
relation (L.2.12), which here becomes
8+r+ xc: 4rn2, (1.3.r5)
we can draw the physical regions for the three channels using oblique
axes as in X'ig. 1.3.2 (Mandelstam, lgbS).

s:0 s:4m2

\
-----------f \ -/-------- t :4m2
,r/

l:0

U:0
Fig. 1.3.2. The Mandelstam diagram using obliquo axes showing the physical regions
(shaded areas) in which s, or t, or z denotes tho squaro of the centre-of-ma,ss onorg.y
for oqual-mass particles in collision.

Since we now have symmetry between the three variables s, t, u, it,

is convenient to change the signs of the four-momenta from those
used in $ 1.2 (equation (1.2.10)) so a,s to give

s : (Pt* Pz)z : (ps* pn)z,)

t:(pr+p)2:(Fz+p)r,l (r.3.16)
u: (pt*p)z : (Fz+p,nY.l
This convention will often be used in the remainder of this book.
A further convention that is sometimes used in the literature is to
write the amplitude l(s,t,z) as a function of three variables, but with
the constraint (1.3.15) relating s, f and u.Ir.fact,,,F is defined only
when (1.3.15) is satisfied so this formal achievement of symmetry is
somewhat ambiguous, and in practice it is better to regard lfl as a
function of two variables, F(s,l) or .F (s, u), for example. Similarly, it
is usually easier to work in the real s, f-plane with orthogonal axes,