To support the discussion of Butler's influence on the series of SPG annual sermons up to the abolition of the slave trade (1807) the entire corpus was examined, the majority of the items were read, and the following searches were performed on page images; the results were tabulated by decade, up to 1800 (after which the culture of the series was disturbed by the campaign for abolition). This appendix gives details of the aims, method and results.

It was hypothesized that a lexical search of the corpus for some of Butler's key words would provide material for a discussion of the influence of Butler's own sermon, and his moral philosophy in general. These were grouped in three categories. After each word the search string is given in parentheses:

1. • Moral philosophy: of the large core vocabulary of the subject, words especially distinctive to Butler were: benevolence (benevolen*); charity (charit*); conscience (conscience); affection (affection*);

2. • Slavery: slavery (slave*); Negro (negro*); Barbados (barbad*);

3. • Law in regard to the master–servant relationship: property (property); family (family); household (household); law (law).

In the last category plurals were excluded from the searches because they would indicate discussion of particulars, not principles.

The word searches were performed on the items held on Gale Publishing's ECCO, using its search engine as it performed on 12–13 October 2009, set to ‘low fuzziness’.

In two cases (Yorke, 1779 and Warren, 1787) ECCO did not hold the sermon and in two others (Harcourt, 1798 and Courtenay, 1800) the images could not be accessed on those days. These four sermons were examined manually at the Bodleian Library, using the same parameters put to Gale's search engine.

The search results were hits on page images; hence two occurrences of ‘conscience’ on one page would result in one hit. The results therefore tabulate not the density of the lexical features but the frequency (per sermon) of pages containing those features. Hence a score of 1.0 indicates that the average sermon used ‘conscience’ on one page. Many hits therefore indicate not greater local density but a more extended consideration of a concept. Variations in purely local density (on a single page, roughly equivalent to an episode in an argument) is an indication of rhetorical style, rather than variations in intensity of concentration on a given concept.