# Biblatex: problem with defernumbers

I'm experiencing a (new) problem with Biblatex after switching computers (for information I am still running the same version 12.04 of Ubuntu but might have a newer version of TeX Live). It seems as though the "defernumbers=true" option used when loading the package does not have the same effect as it used to be. Formerly, after running Biber, three LaTeX runs allowed to reset the numbering of references at 1. It does not work any more. Here is a minimal working example

\documentclass[a4paper]{article}
\usepackage[utf8]{inputenc}
\usepackage[francais,british]{babel}
\usepackage[style=numeric,defernumbers=true,backend=biber]{biblatex}

\begin{document}
\begin{refsection}

The Plemelj-Sochocki theorem \cite{Appel}.

\printbibliography
\end{refsection}

\end{document}

with the contents of Biblio.bib as follows:

@book{
Appel,
author="W. Appel",
title="Mathematics for Physics \& Physicists",
edition="1",
publisher="Princeton University Press",
year="2007"
}

After a single post-Biber run of LaTeX, the reference has label 1 (but things are messy for a realistic document, because the defernumbers option needs two such runs). When I perform the necessary second post-Biber run, the label is raised to 2, and gets stuck there forever regardless of the number of reruns. It is clear from tests with more realistic documents that Biblatex keeps the total number of references in memory and, at the second run, starts the numbering from there. A third run used to correct this on my previous computer. Here is the output:

I hope the issue was stated clearly. Any help greatly appreciated.

• I can verify that you example behaves in this incorrect way with the latest biblatex. May be it is a bug. Mar 4 '14 at 13:19
• I have not seen this issue reported anywhere so your feedback is very valuable. This means the issue is consistent to some degree. Mar 4 '14 at 14:36
• The problem occurs here as well (biblatex 2.8a/biber 1.8), so I suggest you raise this issue with the biblatex bug tracker. Mar 4 '14 at 18:03
• Thanks for the feedback and the suggestion. I reported the problem as advised. Mar 4 '14 at 18:26
• Re-opened, looking into it.
– PLK
Mar 5 '14 at 9:27