We are grateful to Jérémy Tissier, Laetitia Carrive and Eduardo Ascarrunz for fruitful discussions. We also thank Jacques Ducasse who developed modules of the LisBeth program (on which the 3ta analyses were ran) according to our needs, and to Régine Vignes-Lebbe who thought of more efficient algorithms to store and compare sets of 3ts deduced from the 3ta characters. Comments by Mark Wilkinson and an anonymous referees improved this paper. This work was financially supported by the UMR 7207 CR2P.
Received: 4 June 2014
Revised and accepted: 3 February 2015
Published online: 7 July 2015
Editor: M. Brazeau