The vast majority of the quotation evidence in Robert Burchfield’s OED Supplements comes from after the first (1928) edition was completed. The median date for these is 1944, whereas for the first edition it’s 1742. However, in some circumstances the Supplements did reach back into periods already covered by OED1 — if it could antedate an OED1 first citation in a section being revised, for instance.
Or, it appears, in order to bulk up the presence of authors deemed to have been unjustly overlooked by the first edition’s compilers. [hat tip to Danielle for alerting me to a Twain anomaly, which set this post going]
In the two graphs below I show how OED2 supplemented quotations by authors already covered in OED1, separating individual authors from periodical publications. The first, darkest part of the bar, running left from the central zero mark, shows the count of quotations in the first edition (labelled OED1 in the example bar, at top). The next two sections, running right of the zero mark, show the quotations added in by the Supplements, starting with those that could have been included in OED1 (OED2a in the example), based on the publication date, but weren’t, and finally those from works published after the relevant OED1 fascicle, and so could not have been included (OED2b in the example). They are ordered, top to bottom, by the number of quotations that theoretically could have been quoted in OED1, but were not included until the Supplements (i.e., the size of the OED2a segment).
Some helpful details to reading the chart, to start: Knight is the author of technical dictionaries. William James is quoted for his essays and letters. The other ten authors are principally writers of fiction (let’s call plays fiction), which may be surprising, given the relative paucity of fiction in the Supplement. The reason that authors who were long dead before the Supplements have counts in the third segment (OED2-period-only) is that the works being quoted are later editions of earlier works, e.g. George Eliot’s Letters (written between 1840 and 1880; but in the published in various editions and collections from 1880, 1885, 1932, 1954, 1956).
What the chart suggests is that literary works were more likely to be re-examined by the Supplement than other kinds of texts. But actually this is unreliable, and hardly a surprising take-away, since although literary quotations do not make up a large proportion of quotations overall, they are drawn from relatively few works, meaning that the top individual authors are all literary authors. (They are, in the Supplements: Shaw, Joyce, Kipling, Wodehouse, Lawrence, Twain, and Huxley [this excludes multiple-authored works such as newspapers, which best all of these names in the Supplement]). The same principle principle applies to OED1 sources.
There are another few things to observe about this chart. The first is that several of the authors on the top end were flourishing during the time period in which OED1 was being published (1884-1928). Burchfield himself makes a comment related to this, mentioning one of the names on our list:
The language of Thackeray, Swinburne, Henry James, and others had been too uncomfortably close in time for Murray and his colleagues to take it fully into account (Burchfield, Unlocking the Language, 173)
Actually it looks like Thackeray was fairly well covered by OED1, relative to OED2’s additions. But the basic point is probably right–the nearer a source to the time of preparation of the dictionary, the less likely it is to be incorporated, all other things being equal. Shaw won the Nobel Prize in 1925, and enjoyed success in the culture from at least the mid-1890s, but A) many of his works were not published until much later, and B) his early-published works do not seem to have been very much used by OED1.
Twain, who has the most OED1-eligible quotations added back by the Supplements, is interesting, partially because he is not one of those authors routinely mentioned by Burchfield as being of special literary interest to the dictionary (these tend to be some combination of Joyce, Lawrence, Woolf, and Kipling). A clue lies in the publication of A Mark Twain Lexicon in 1938, which Burchfield reports was a key source for the Supplement (in “Data Collecting and Research“, 1973). The effect is of course smaller than those heavily concorded corpora of the early modern period, but it is instructive nonetheless: again we see the value of word lists to the lexicographer, especially when they refer back to quotable texts.
It’s also telling, I think, to find three women among these authors. This would tend to support the idea that the Supplement paid more attention to female authors than OED1, but I’ve recently cast some doubt on that conclusion (see Sex in the OED).
Here’s the same graph for periodicals, which I’ll leave without much comment. We already knew that the Supplements made much greater use of periodical literature than OED1. This shows that at least some of that involved going over material that would have been available to the original editors (amongst whom the issue of how and whether to use newspaper quotations was a major bone of contention. See discussions in Gilliver, The Making of the OED).