1: \begin{abstract}

2: The consequences of adopting other definitions of the concepts of sum and

3: convergence of a series are discussed in the light of historical and

4: epistemological contexts.  We show that some divergent series appearing in the

5: context of renormalization methods cannot be assigned finite values in a form

6: consistent with Hardy's axioms without at the same time equating one to zero,

7: thus destroying the mathematical building. We show that if the replacements for

8: the concept of sum of a series are required to be associative, to be invariant

9: under finite permutations of the terms and dilution, further restrictions

10: emerge. We finally discuss the epistemological costs of accepting these

11: practices in the name of instrumentalism.

12: \end{abstract}

13: