For every real t one has
∣Γ(it)∣=tsinhπtπ(t=0)
∣Γ(21+it)∣=coshπtπ.
Indeed, one deduces from (8) that Γ(it)Γ(−it)=tsinπitiπ=tsinhπtπ, and one has Γ(−it)=Γ(it); similarly, (8) gives
Γ(21+it)Γ(21−it)=sin(2π+πit)π=cosπitπ=coshπtπ,
and one has
Γ(21−it)=Γ(21+it).
Now let p be any integer >0 and consider the product
f(z)=Γ(pz+1)Γ(pz+2)…Γ(pz+p).
By (3) (VII, p. 315), for every z=−n (n∈N), f(z) is the limit of the product
(pz+1)(pz+1+1)…(pz+1+n)n(z+1)/pn!
(pz+2)(pz+2+1)…(pz+2+n)n(z+2)/pn!⋯
…(pz+p)(pz+p+1)…(pz+p+n)n(z+p)/pn!
=(z+1)(z+2)…(z+(n+1)p)nz+(p+1)/2p(n+1)p(n!)p
and in particular f(0) is the limit of the product
((n+1)p)!n(p+1)/2p(n+1)p(n!)p
from which it follows that f(z)/f(0) is the limit of
(z+1)(z+2)…(z+(n+1)p)nz((n+1)p)!=z p−z(n+1n)z⋅z(z+1)(z+2)…(z+(n+1)p)((n+1)p)z((n+1)p)! which, by (3) (VII, p. 315), gives
f(z)=f(0) z p−z Γ(z). But one can write
f(0)=k=1∏p−1Γ(pk)=k=1∏p−1Γ(1−pk)=k=1∏p−1Γ(pk)Γ(1−pk) since
f(0)>0; the complements’ relation then gives
f(0)=πp−1/k=1∏p−1sinpkπ and since the product on the right-hand side is equal to
p/2p−1 (VI, p. 284, cor. 1) one sees finally that:
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.
Nothing.