Original official author HTML, exact retained edition. Historical TeX conversion verdicts remain unchanged. Cited-edition alignment and mathematical self-containment are not assessed.
Let be an open affine cover of and set . Then is strictly -affinoid by [Bos77, Theorem 3.1] and hence is a good germ. Note that is a cover of . Hence is obtained by glueing the along the canonical maps . Let for a strictly -affinoid algebra . Then and the character induces a morphism . Let be the prime ideal corresponding to i.e. is the kernel of . The induced morphism is injective and hence it extends to a morphism where denotes the function field of . This induces a morphism .
First step: We have that is the preimage under of the set of valuation rings in which admit a center on .
Indeed if is a valuation ring with then . Let be the maximal ideal of then defines a point in whose local ring is and we have . Then admits the center on as claimed. Conversely if admits a center on then there exists such that and hence obviously .
Second step: The map is surjective if and only if any valuation on admits at least one center on .
Let be surjective and a valuation on . Then extends to a valuation on . Let be the valuation ring of . Then and hence has a preimage . Then there exists such that hence the image of in admits a center on by the first step. But this image is by construction which is the valuation ring of . Hence admits a center on . Conversely suppose that any valuation on admits a center on and let then the image of in induces a valuation on which admits a center on . Let such that then and the induced element in is a preimage of .
Third step: The map is injective if and only if every valuation on admits at most one center on .
To see this we describe . In order to do so we cover by open affine subsets . Their preimages under yield a cover of by good germs. As above their reductions can be described as the preimage of the set of valuation rings in which admit a center on . The reduction of is then obtained by glueing these spaces. Now suppose that any valuation on admits at most one center and let which map to the same valuation ring . There exists such that . As we have seen in the first step, and admit centers respectively . Then both are a center of . Hence by our assumption. Therefore by the first step in . Hence in the glueing process, and are identified with each other. Conversely suppose that there is a valuation on which admits two centers . Let and . Choose an extension of the valuation to and let denote its valuation ring. Then induces an element as well as an element . Then and map to the same element in but they are not identified in the glueing process as and are separated and hence and admit at most one center in respectively which means in particular that they do not admit a center in . Hence is not injective. This proves the third step.
Recall that is proper if and only if every valuation on admits a unique center on ([Har77, Ch. II, Ex. 4.5]). Hence the claim follows from the second and third step.
∎