The Google Ngram Viewer visualizes phrase and word usage across the 5.2 million books digitized by Google up to 2012. Results are normalized by the number of books published in each year, and smoothed using a moving average of three years. The chart for ‘library classification’ shows a worrying trend.

ngram (library classification)

It makes sense that interest in classification, as reflected in Google’s corpus, peaked in the 1960s. Many American research libraries converted to LCC at this time. The decade was also the heyday of Classification Research Group, based in London, and the Classification Research Study Group, led by Phyllis Richmond.

But have yesterday’s questions about classification really been settled? With most public and school libraries using Dewey, and academic and research libraries using LCC, is there nothing more to talk about? Have online tools such as Classification Web, WebDewey, and OCLC’s Classify, replaced the need for classification training in library pedagogy, and left communities of practice with little to discuss?

If so, we’re becoming dangerously complacent. LC does a superb job of maintaining LCC, but as I’ve argued elsewhere, the scheme suffers from systematic bias. Dewey is even worse – it maintains the prejudices of its creator, and the overconfidence of a precocious young man straight out of his undergraduate degree.

Dewey is so hopelessly out of date, and unnecessarily overcomplicated, that viable alternatives are now being implemented. But how can we reclassify if we’ve stopped talking about classification? And how can we classify if we don’t even know what classification is?


Phyllis Allen Richmond (1921-1997) was one of the greatest 20th-century classificationists, and I return to her writings whenever I despair of the obscurity and narrow focus general among many modern writers in the field. I share her skepticism of the library classification schemes that attempt to make the world fit an inflexible model, however granular or faceted that model might be.

Her essay, ‘Some aspects of basic research in classification’ published in Library Resources and Technical Services in the Spring of 1960, takes some well-aimed potshots at the cult of Ranganathian subject analysis, then prevalent. Here’s a brief extract:

‘The Dewey, Universal Decimal and Bliss classifications are made primarily by logical division, which is mirrored to some extent in their notations. The Library of Congress classification, in contrast, is definitely non-logical and the notation is largely ordinal. This classification functions as well, if not better, than the others. A non-logical classification has the great advantage of flexibility, since one may add to it rather freely without upsetting the whole pattern. The fact that the Library of Congress classification is displayed in an almost random fashion, after some initial form divisions in each main class, does not seem to be a disadvantage in its operation. In a non-conventional classification, the elements of each category may be expressed either in a logical or a non-logical manner. The advantage of variability, rather than the display of relationships, is the chief motivation for choosing a non-logical arrangement.’