The geometry of Niggli reduction: BGAOL – embedding Niggli reduction and analysis of boundaries. Erratum

Ambiguities in the article by Andrews & Bernstein [J. Appl. Cryst. (2014), 47, 346–359] are clarified.

In discussions with R. Oishi-Tomiyasu, we have realized that some statements about the work of Oishi-Tomiyasu (2012) in our recent paper (Andrews & Bernstein, 2014) may have given some unwarranted impressions.

Our discussion of the linearity of metrics was mainly meant to address issues that arise when a metric is implemented for database searches. For the case of Bravais lattice determination, most of the trial positions are already inside the Niggli code or essentially on its surface. We should have said that the discussion of linearity is not important in this context for Oishi-Tomiyasu's codes.

In the discussion of the depth of iteration required, we focused on the fact that the searches require more depth in certain regions (especially face-centered cubic, for example). We should point out that although Oishi-Tomiyasu's (2012) implementation requires slightly more computation in that region, it requires significantly less than the extensive geometric searches implemented by earlier authors, particularly because it already understands the few regions that must be searched.