By Eqs. (4-155), (4-157) and (4-170)



 Qa[e-1] ~ (sin 4)" (cos ))2J X-(p/2+j)As-- 0, (4-198)



 where s = -3 for a t, r and s = -2 for a = .

 (ii-2) n q 0
 By Eqs. (4-155), (4-157) and (4-171)



 Qa [e-] ~ (sin )" p(cos 4)2j X-(p2+j)Rs, (4-199)



 where s = -2 for a t, r and s -1 for a = 0, .

Therefore, by analyzing the structure of Qa[-'1] we find that the e-1 terms vanish

in all the cases except when n q = 0. The non-vanishing Ba-terms are derived

only from this case. Then, by 0 < q < 1p + j (p + i), 0 < i < k p and
p < k < 2n together with n = q one can show that

 p+i < kandk <p+i, i.e. k p+i. (4-200)

Substituting this result into Eq. (4-190), it is confirmed that in the numerator of

Q[e-1] in Eq. (4-188) one can simply substitute

 ( -_ o)k-p ( ')k-p Q. E. D. (4-201)

 The significance of this proof does not lie in the result given by Eq. (4-201)
only, but also in the fact that the non-vanishing contribution comes only from the

case n = q for Eq. (4-197). By 0 < q < +p + j (p + i), 0 < i < k p, p < k < 2n
and n = q we find k = 2n and then 2j = 2n p. Therefore, Eq. (4-197) finally