Finding Early Modern Bee Books: The Sources
British Bee Books: A Bibliography
What constitutes a book about Bees? What is in the scope of this project and what is not? Sources are the bread and butter for the historian. These can take all kinds of shape and form: texts, images, videos, archaeological finds, and so on. The most common is, of course, texts. These fall into various categories, but mainly into primary and secondary material. The core corpus for Early Modern Bees are the English publications from the earliest printed texts in the late fifteenth-century to those produced around 1700. I’m interested in any text that talks about the Honey Bee, which, it appears, would number in the thousands.
It is not all that long ago that such a task was virtually impossible. Using archives and libraries I could have scanned through the masses of printed literature to search for references, but this would have been cumbersome, inefficient, and time-consuming. It would have certainly been impossible to claim any kind of comprehensiveness in the research, the best that I could have achieved was a wide-spread of evidence; the result would have been more an indication of the available corpus or a selection based on specific criteria, rather than anything complete. The best that could be achieved on such a pursuit has most likely been done. In 1979, Joan P. Harding, David A. Smith, Eva Crane, and Rosamund M. Duruz, published British Bee Books: A Bibliography, 1500-1976. This, in their own words, ‘took twenty-five years of preparation,’ with ‘entries compiled by many people’. The scope of that work was to list every book that contained more than 20 pages on bees, published in Great Britain and Ireland up to 1976. For our period, some additions break that rule:
to complete this picture, particularly for the earlier years, other books with references to bees are included – herbals, books on agriculture, books on insects – and we have upon occasion broken the rule and included works of less than 20 pages, or more general works with less than 20 pages dealing with bees. (Smith, 1979, 12).
British Bee Books identifies 74 items, which were published between 1500-1700, 37 of which might further be described as ‘how-to’ books (providing guidance in full or in part as to how to work with bees). Most works of fiction are avoided (although some full-sized fictional books that ‘can reasonably be described as bee books’ are included), only the most important classical reproductions are listed (e.g. the best-known translations of Virgil’s Georgics), articles in periodicals are listed up to 1800. This is an extremely useful start and the British Bee Books does indeed achieve its main purpose of being comprehensive in terms of books where the main (or significant) focus are honey bees.
The NGram Viewer
Via the Google NGram viewer, which reveals some of the statistics behind the mass corpus of books that Google have digitised, it is possible to identify some indication of relative interest in a topic. Thus, for ‘Bees’ the NGram viewer (set with a smoothing of ‘0’*) suggests that there were spikes of publications that contain the word ‘Bees’ in the latter half of the sixteenth-century (the 1580s onwards), again between 1620 and 1640, and then again around 1660, with quite a considerable drop in the tail-end of the seventeenth-century.
What does this mean? For this period the scans of books are less comprehensive than, for instance, the eighteenth- and nineteenth-centuries. Thus, once again we cannot claim comprehensiveness from this data, nor complete accuracy. The spikes could just as easily tell us something about Google’s priorities as it might about actual interest in a topic in any given period. It might also be affected by bad scanning (or more precisely bad OCR – the method for making a text readable by computers) and it does not consider variant spellings. However, the NGram viewer does enable us to add to the picture initially supplied by British Bee Books.
Number of books: NGram Viewer
Number of books: British Bee Books
Table comparing entries in British Bee Books (1979) and the NGram Viewer [June 2018]
The table works on the date ranges supplied by the Google NGram Viewer on 25 June 2018. On the surface, it would appear to suggest that in all but one instance the number of books listed is more complete in the British Bee Books, than on Google Books. The one exception, the date range 1687-1700, seems to have a variance of 5 books, in Google’s favour. What is interesting about all of this is that very few of the books listed in the British Bee Books are to be found in Google’s dataset (and vice-a-versa). For the 1687-1700 date range, none of the books are the same.
1687-1700 Date Range
Google Books NGram Viewer
British Bee Books
· W.R., A Necessary family book
· John Aubrey, Natural History of Wiltshire
· Thomas Tryon, The Good housewife made a doctor
· William Y-worth, The Britannian magazine
· Joseoph W. Ebsworth, Merry Drolley Compleate: Being Jovial Poems, Merry Songs, &c.
· William De Britaine, Humane Prudence: Or the Art by which a Man may raise himself
· The Genuine epistles of the Apostolic Fathers
· Sir Thomas Pope Blount, A Natural History
· Baltasar Gracián y Morales, The Courtier’s Oracle: Or, The Art of Prudence
· The Office of Consecrating Cramp Rings
· The Whole Book of Psalms
· Gervase Markham, A Way to Get Wealth
· John Milton, A Defence of the People of England (x2)
· Art’s treasury of rarities and curious inventions
· Sir Simon Degge, The Parson’s Counsellor
· Richard Allestree, The Whole Duty of Mourning
· George Hartman, The Family Physician
· Virgil, The Works of Virgil
Table showing books listed on the NGram Viewer with search term ‘bees’ (June 2018) and British Bee Books (1979) for the date range 1687-1700
The NGram viewer has therefore added to this date range twelve new items (actually eleven, as John Milton’s A Defence of the People of England is listed twice) that might be worth examining.
Of the 1500-1583 date range, three books are the same between the two datasets: Conrad Heresbach’s Four Books of Husbandry; Thomas Hill’s The Profitable Art of Gardening; and Isaac Rabotenu’s The Bee Hive of the Romish Church (the latter of which also appears as a later edition in the 1593-1686 date range in the NGram viewer). Three others, therefore, can be added to the corpus. In the 1593-1686 date range, only four books agree: Charles Butler’s The Feminine Monarchie; Samuel Hartlib’s The Reformed Common-wealth of Bees; John Evelyn’s Discourse of Bees; and A Full Discovery of Bees, which is the second edition of Moses Rusden’s A Further Discovery of Bees (printed in 1679). Nineteen additional books can, therefore, be added.
However, there is a cause to doubt all of these statistics. Whilst the NGram viewer itself does not seem to alter, the individual results sometimes can. For instance, the 24 items listed for 1593-1686 appear across 3 tabs with ten items shown per page, but sometimes this number jumps to 8 or 13 tabs (but does not always show all those results). Similarly, the 12 items listed between 1687-1700 take 2 tabs but can jump to 8-9 tabs. Thus, the data is partial, and attempts to delve further into the detail reveals inconsistencies. Nonetheless, even if the statistics are dubious, as a finding tool Google Books remains useful.
Project Gutenberg and the Internet Archive
It appears to be impossible to search Project Gutenberg and the Internet Archive by date of publication (you need to know what you’re looking for), so whilst these are useful for finding specific texts, they are not good for identifying a corpus.
Early English Books Online
The EEBO-TCP (Early English Books Online Text Creation Partnership) full-text search (which usefully ends at 1700) brings up 14,842 matches in 4,374 records, whilst a search on Jisc Historical Books (a subscription version of a similar database) brings up 4,206 items (and 6 additional undated items). It is hard to identify the reason for this discrepancy as both databases are based on the same list of 125,000 books, and borrow from one-another (a difference in the search engine or the metadata itself?). Historical Books claims that as of June 2016, 40% of their database includes full-text searchable text, OCR’d by the human eye, whilst EEBO-TCP claims over 50,000 books (over half of which are available for free). The project is split into two phases EEBO-TCP Phase I, all of which have been released under a public licence, and EEBO-TCP Phase II, which is available only under subscription at present but will be released under a public licence in 2020. Taken together, these databases are, by far, the nearest to comprehensive we are going to get.
It is not perfect. Whilst EEBO/Historical Texts boasts as complete a collection of early modern printed texts as its possible to get, it does not yet offer them all in machine-readable formats (less than half). Not all publications are full-text searchable, many items rely entirely on metadata to be found in a search term. Despite these caveats, EEBO significantly adds to the identifiable corpus of Bee Books and has the potential to offer more useful statistics.
Figure: Search term ‘bees’ between date ranges 1450-1700 [Historical Texts: June 2018]
As the graph shows, there are spikes of interest around 1600, in the 1660s and again around 1680 (which is not too far off from the statistics given by the NGram viewer). The trend is one of a gradual increase, in part because more publications come onto the market as the seventeenth-century develops.
A final caveat
Using the term ‘bees’ to search the various databases for books on the honey bee, is not without problems. It does not account for other languages (particularly Latin) and it neglects the singular ‘bee’. Of the former, searches under ‘apes’ (2396 results) or ‘apis’ (553 results)** might bare useful results. Of the latter, the singular term fails to capture useful data. Too many early modern books spell ‘be’ (as in ‘it might be a long night’) as ‘bee’. It is also worth noting that even the term ‘bees’ is not always accurate. In this respect, the village of St Bees in Cumbria is a real nuisance (sorry St Bees!).
Relying on the term ‘bees’ also does not account for texts that mention only honey or beeswax. Whilst searches can be conducted on both words, there are more difficulties ahead. There are a variety of spellings for honey in this period (huny, hony, hoony, honny, hiney)***, as the graph from the Google NGram viewer below indicates. ‘Honey’ and ‘Hony’ are the most popular, but the other variants did exist. A search for ‘honey’ therefore requires multiple or variant searches.
Meanwhile, beeswax (which should be captured in a search for ‘bees’ if the term is used in full) is often just referred to as ‘wax’. In this period there are two main types of wax in production, tallow, and beeswax, and it is quite a task to distinguish between them when handling a large corpus such as this.
The search is also incomplete because the corpus is incomplete. Not all publications from the period survive in full or in part. Ephemera (pamphlets, etc.), especially, have tended to disappear. The result, therefore, is always going to be slightly skewed. Additionally, and as referred to at various points already, digitisation, especially machine-readable digitisation, and thus 100% accurate searches, are far from perfect at present. Certain items might be missed either because they have not yet been made machine-readable or because of errors in the transcription. It is for this reason that I plan not just to rely on the EEBO databases, but also on other search engines, such as Google Books, Project Gutenberg, and the Internet Archive.
The corpus will, therefore, never be entirely complete. Perhaps this does not matter. A corpus of just over 4,000 books would seem to at least offer as near to comprehensive as possible, whilst accepting some losses and inaccuracies. It is certainly more than we could achieve just thirty or twenty years ago and justifiably enough to offer useful insights into the corpus, once further research has been conducted.
Joan P. Harding, David A. Smith, Eva Crane, Rosamund M. Duruz, British Bee Books: A Bibliography, 1500-1976 (International Bee Research Association: London, 1979)
* A smoothing of 1 means that the data will include the average of the raw count for one year plus one year either side, divided by 3. This helps to make the data more legible where big data is involved. However, for the period 1500-1700 there are fewer data and thus a smoothing of ‘0’, i.e. raw data, can be more useful.
** Data from Jisc Historical Texts [June 2018]
*** Taken from a search of ‘honey, n. and adj.’ from Oxford English Dictionary Online [June 2018]