Ooops. ‘Figures’ pages were being indexed in JURN for the verbose biomedical full-text website ncbi.nlm.nih.gov. Now they’re not, as I’ve excluded them.
Not that it mattered all that much, as they were being automatically suppressed by Google’s de-duplication and results ranking. In some cases Google appears to avoid ncbi.nlm.nih.gov altogether in favour of full-text at a better source repository. For instance, search JURN for the URL https://www.ncbi.nlm.nih.gov/pmc/articles/pmc3566601/ and find a link to the article at NCBI, but search for the same article’s title and Google ranks its MIT repository location as the first result — and ignores NCBI because it’s deemed a duplicate of the more worthy MIT.