Advertising Pornography is a package of two complementary datasets derived from advertising material issued by major figures in the early Victorian pornography trade, including William Dugdale, Edward Duncombe, John Duncombe, Edward Dyer, and Thomas Paine Carlile. One of the datasets, Sales Catalogues, 1840-1856, is derived from 16 unique sales catalogues and two duplicates with marginalia, and records more than 1600 items listed for sale. The other dataset, Periodical Advertising, 1822-1870, is derived from 581 unique periodical advertisements and records more than 5000 items listed for sale. Businesses oriented around trade in sexual material began to emerge in London in the 1820s. Amid the decline of revolutionary politics, a number of radical pressmen shifted their focus to this material, establishing themselves in Holywell Street and other places near London’s Strand.1 The period between 1840 and 1856 was their heyday, coming after traffic in sexual material crystallized as a specialized sector of the marketplace and before the introduction of mass-market photographs substantially reoriented it.
The data’s cultural significance extends beyond its relevance to the history of pornography, offering insights into the developments in media culture, popular literature, art, music, and theatre, radical politics, and reproductive health. Advertising Pornography illustrates that early Victorian pornographers did not only sell expensive novels, demonstrating the important role that visual material played in the trade and producers’ rapid adoption of new visual media technologies. It also illustrates that many of the items sold in the trade were inexpensive, and that a wide variety of items — including sex toys, condoms, paper dolls, playing cards, tobacco pouches, song and joke books, translations of popular novels, guides to London’s night life, and medical works — were advertised for sale alongside graphically explicit literature and images. The datasets facilitate analyses of inventory, prices, and item descriptions, make relationships between players and their geographic movements more legible, and make it easier to understand what bibliographical studies have captured of this print culture and what they have missed. Advertising Pornography also makes it easier to simply explore what kinds of sexually related items were sold during the nineteenth century, and how particular items were priced, described, and distributed in this context.
Background and Context
Studying the development of pornography and the people who made and sold it requires working with a fragmentary and often untrustworthy historical record.2 In nineteenth-century Britain, sellers of sexually explicit literature and images were routinely prosecuted under obscenity laws. These actions produced court, government, and news records that have helped scholars understand how the pornography industry emerged, what early specialists in sexual material sold, and how they operated their businesses. However, they also encouraged producers and distributors to cover their tracks. These figures’ routine use of aliases, false imprints and publication dates, and “front men” to move their wares has made establishing exactly who made and sold what when challenging.3 The fact that relatively little of the trade’s output survives has exacerbated these challenges. Prosecutions increasingly culminated in the mass destruction of stock. Items that survived destruction were rarely preserved for posterity. Some were purchased by wealthy collectors and ended up in research library collections. These collections are priceless resources. However, since they were shaped by collectors’ personal interests and networks as well as librarians’ judgement about what items were worthy of preservation, they don’t offer a reliable index of what was made and sold in the trade.4
Scholars address some of these issues by casting wide nets, piecing pornography’s history together by compiling and analyzing fragments of evidence from documents in many physical and digital collections.5 My main goal in creating Advertising Pornography was to facilitate that work by making it easier to search and cross-reference information from advertising material issued by agents connected with the trade. The datasets have been designed so they can be used separately or together to compensate for drawbacks associated with using different kinds of advertising material as sources. Sales catalogues often contain rich descriptions of the items that pornographers sold, and occasionally marginalia. However, few catalogues have survived. Most are also undated and were published anonymously or under pseudonyms. Some do not even provide an address for the seller. In contrast, periodical advertisements survive in huge numbers and have clear dates of publication. Although they were frequently published under pseudonyms, they virtually always include an address, which makes it easier to identify sellers. However, periodical advertisements offer less detail about the items being promoted than sales catalogues do. Often, the only information that they provide is a short title and a price. Moreover, pornographers did not advertise all the items they sold in periodicals. Together, the datasets furnish a clearer picture of what they sold and how and when they sold it than they do individually.
The datasets also facilitate research by indicating whether each item listed for sale in advertising material is mentioned in bibliographies that historians of English pornography often rely on: Peter Mendes’s Clandestine Erotic Fiction in English, 1800–1930 (1993) and Henry Spencer Ashbee’s Index Librorum Prohibitorum (1877), Centuria Librorum Absconditorum (1879) and Catena Librorum Tacendorum (1885).6 These bibliographies offer details about some items that are not available in the advertising material. At the same time, the advertising material fills in details about some items unavailable in these bibliographies and helps identify gaps in publication histories they offer. The value of using these sources together is demonstrated to some degree through a supplementary document included with Sales Catalogues, Catalogue Dating and Attribution. This document explains the working publication dates and attributions that I assigned to each of the catalogues, which are grounded in research that relied partly on data from Periodical Advertising and partly on bibliographies*.*
Collection and Creation
Sales Catalogues, 1840-1856
This dataset is derived from eighteen sales catalogues issued by agents associated with the British pornography trade. Fourteen of the catalogues are from the Michael Sadleir Collection of Ephemera, 1748-1887 at the Lilly Library, Indiana University. Ten of these catalogues are “standalone” catalogues, that is, catalogues issued for distribution on their own, often by mail-order. Two were published in numbers of William Dugdale’s periodical The Exquisite, and two are bound in books. The rest of the catalogues that inform the dataset are bound in books held at the Kinsey Institute Library and Special Collections, Indiana University (1 catalogue), the Harry Ransom Center (1 catalogue), the Wellcome Library (1 catalogue), and the British Library (1 catalogue). I acquired scans of these catalogues during research for a previous project.
To create the dataset, I prepared an Excel file with two sheets to contain catalogue data and metadata. The fields in each sheet are described in the README. I then 1) transcribed the text of the catalogues, 2) checked the transcriptions against the original scans, 3) pasted the transcribed text into the appropriate fields in the Excel sheets, 4) filled in the metadata fields, and 5) did a second check of the transcribed text for accuracy. I created transcription drafts for eight of the catalogues using automated text recognition via Claude 3.7 Sonnet, primarily because I was interested in getting a sense of the model’s ability to accurately transcribe text from historical documents that were unlikely to be in its training data at the time. I uploaded scans of these catalogues to Claude one by one and prompted it to extract text from each catalogue exactly as written. This was largely successful. However, the model occasionally skipped over content in the catalogues or hallucinated words or phrases (it was most prone to do this when attempting to transcribe passages containing Victorian slang). To ensure the dataset’s accuracy, I hand checked the transcriptions against the original scans twice, once after they were generated and once after I had entered them into the spreadsheet, correcting the text as needed. I found that the type in the other catalogues was too small or blurred for Claude or specialized text recognition software to transcribe with an acceptable level of accuracy. I transcribed the text of the remaining ten catalogues entirely by hand. I checked my own transcriptions against the original scans twice.
Some of the catalogues are riddled with misspellings and typographical errors. Misspellings are especially common for French titles and French-language item descriptions, which appear in two of the catalogues. I decided not to correct these errors as they offer historical evidence of the conditions in which the catalogues were issued. In some cases, I placed a [sic] next to misspelled words or typographical errors, and in some cases (especially for very garbled French) I indicated the likely proper spelling of words or phrases in square brackets. However, I did not flag all misspellings with these devices as that would make the data difficult to parse (e.g. I did not flag French words with missing accents since they are so numerous in the French-language titles and item descriptions).
In some cases, classifying items listed for sale in the catalogues was challenging. It was impossible to tell whether many items were cheap books or pamphlets, so I ultimately labelled all bookish items “books” in the metadata field focused on physical format.7 When classifying items by genre, I followed the way the items were represented in the catalogues. For instance, if an item appears under the header “medical books” in one or more of the catalogues, I categorized it as “medical” even if twenty-first century readers would not necessarily consider it a medical work. This approach resulted in a very broad “literary” genre category, encompassing novels, poetry, biographies, and essays. In a few cases, the catalogues did not give a clear impression of an item’s genre, and the item also does not appear to have survived, so I could not rely on my own judgement. In these cases, I designated the genre “unknown.” In a few cases, it was hard to tell whether an item listed for sale was a book or a print collection. In those cases, I assigned it with the most likely format label and noted my uncertainty with a question mark or a note attached to the item entry.
Periodical Advertising, 1822-1870
This dataset is derived from 581 periodical advertisements issued between 1822 and 1870 by figures who issued the catalogues in the Sales Catalogues dataset, their associates, and their competitors. Except for two advertisements, which appeared in the periodical Sportsman's Magazine of Life in London and the Country (London: E. Dipple, 1845), which has been digitized by Google, all of the advertisements in this dataset were published in periodicals that have been digitized and made accessible through the British Newspaper Archive (BNA).
To create the dataset, I prepared an Excel file with two sheets to contain advertisement data and metadata. The fields in each sheet are described in the README. Unfortunately, the BNA does not have an API or otherwise offer access to the text of all of the newspapers in the collection in bulk.8 I therefore used the BNA’s keyword search function to search for relevant advertisements using names, titles, addresses, and terms that I knew from my previous research on the trade and from Sales Catalogues were likely to appear in them. I transcribed data directly from the scans in the BNA into the Excel sheets entry by entry and filled in the metadata fields as I went. When filling in metadata fields that also appear in Sales Catalogues, I used the same controlled vocabularies I used in the Sales Catalogues dataset to make it possible for users to work across the datasets.
I aimed to record one example of every unique advertisement that every agent associated with catalogues in Sales Catalogues placed in every year they advertised up to 1870.9 Although I had previously worked with these agents’ advertisements, finding them still involved a lot of trial and error. The garbled state of the machine-readable text in the BNA meant that relevant search terms did not always surface advertisements in which they appeared*.* In some cases, I skimmed through periodicals likely to publish relevant advertisements to get around this issue. During the collection process, I found advertisements issued by other figures active in the trade. I recorded their advertisements too, aiming to record one example of every unique advertisement placed by such figures in every year they advertised up to 1870.
I took the same approach to spelling and typographical errors and classification when creating this dataset that I used for Sales Catalogues. Because the periodical advertisements provide much less information about items listed for sale than the catalogues do, there are more cases in which I was unable to establish the genre of an item, know for sure whether an item was a print collection or a book, or know whether similar titles were variant titles of the same work. As in Sales Catalogues, I noted such issues beside the relevant items in the ‘notes’ field and/or indicated uncertainty with a question mark.
Post-Production Editing
After I completed the datasets, I checked metadata fields in each for typos and compared entries in fields with controlled vocabularies to ensure consistent use across the datasets.
How Researchers Might Use These Datasets
These datasets may prove useful to researchers working in a variety of different areas, including Victorian studies, literary, art, music and theatre history, book history, media studies, the histories of medicine, gender and sexuality, business history, and legal history.
I anticipate that some researchers may use these datasets like databases, using word searches and field filters to examine listings for items or kinds of items in the advertising material. If you’re studying a specific title, such as the popular midwifery manual Aristotle’s Masterpiece, you can filter for it using the Short_Title column in each dataset and see how the work was priced, who advertised it for sale and when, and how it was described. If you’re interested in trade in certain types of items (such as sex toys or photographs) or in particular genres (such as song books or medical books), you can use the Item_Type and Genre columns to filter for items in these categories in each dataset.
Other researchers may prefer to use the datasets to generate statistical information about pricing, format, or genre, or analyze the data in other ways. To give a handful of straightforward examples: you could use the datasets to analyze pricing patterns for Aristotle’s Masterpiece, or medical material, or books generally, in individual catalogues, across mail-order catalogues, across a single player’s catalogues and periodical advertisements, or across all of the advertising material represented in the datasets. Moreover, you could analyze pricing in a single year, across the entire period covered in the datasets, or some period in between. The datasets also enable studies of inventory flows, sellers’ relationships and geographic movements, and their use of descriptive language. Such work could help us better understand how the trade operated and changed over time.
The datasets could also be adapted or used as a basis for other digital projects. Some scholars may wish that they came with different metadata. If this is you, feel free to edit it. Others may be dismayed by the fact that Periodical Advertising ends in 1870; require a random sample of advertisements; or want to answer research questions for which information about duplicate advertisements is crucial (see notes on limitations below). If this is you, you could use these datasets as starting points to help you find more advertisements and build a dataset that works for your needs. Finally, the datasets may prove useful in digital humanities teaching. Paring them down (e.g. deleting some metadata columns) may make them more suitable for some teaching applications.
Limitations
Advertising Pornography does not offer a complete or unmediated picture of the past. Sales Catalogues represents what has survived—and of that, what I have located—of the sales catalogues that agents associated with the early Victorian pornography trade issued. Given the content of their periodical advertisements and findings from prior studies, I believe that the dataset offers quite an accurate illustration of what was made and sold in the trade. However, we don't know what was in catalogues that did not survive. Several catalogues from the Lilly Library collection also appear to have been annotated by the same person, raising the possibility of bias introduced by a single customer's habits.
Similarly, Periodical Advertising is derived from a curated collection of advertisements whose creation was shaped by the selection of periodicals that the BNA has included in the archive to date, my reliance on known terminology for keyword searching, and search failures caused by OCR errors. Although I aimed to construct a representative picture of the content of early nineteenth-century pornographers’ advertising in periodicals from year to year, these issues limit claims we can make about the data. Future mining of the BNA—especially with improved OCR and/or identification techniques that do not rely on keywords—or of periodicals not included in the BNA will likely reveal new advertisements.
The fact that the dataset does not offer information about duplicate advertisements also limits its uses. For instance, it means that Periodical Advertising does not offer complete or representative information about where pornographers advertised.10 It also means that the dataset is unsuitable for estimating different agents’ spending on advertising or for measuring how frequently some agents advertised relative to others.
Some of the information in both datasets is also unreliable because the people who made the material that informs them were unreliable. Victorian pornographers’ claims about the authorship and provenance of the works they sold are often (but not always) false. For instance, the work The Secrets of Nature Revealed is not “a translation from the Latin of Michael Scotus, the famous philosopher,” as several catalogues claim, but a compilation of chapters extracted from two popular early modern medical works.11 Pornographers’ claims about the quality of the items they sold should also be taken with a grain of salt: illustrations described in effusive terms in advertisements can turn out to be pretty shoddy pieces of work.12 As Lisa Z. Sigel has put it, working with nineteenth-century pornographers’ output requires an attitude of “radical skepticism.”13 Basic claims about a book or print can be lies—and claims that sound like lies can be true.
Finally, it’s important to be aware that like many other nineteenth-century book trade workers, pornographers reprinted and reworked a lot of old content. Listings for different editions of the same works are grouped together in the short title fields in both datasets. The material design and content of these editions could be quite different, however. For instance, non-explicit and pornographic editions of The Confessions of Madame Vestris were issued during the nineteenth century and appear to have circulated simultaneously.14 It’s especially important to keep this in mind when looking at listings for titles that were first published before the nineteenth century: a literary work that you know from eighteenth-century print culture may have been edited substantially.15
Ethical Considerations
Catherine D’Ignazio and Lauren Klein have emphasised that decisions about data collection and presentation are never neutral: they can reflect and reinforce — or expose and disturb — structures of power.16 Creating these datasets required that I consider two separate issues involving potential harm.
First, I needed to decide whether to include offensive language, which often appears in historical texts. A small number of item titles and descriptions in the sales catalogues and periodical advertisements that inform these datasets use crude terms for body parts and sexual acts or refer to sexual assault. A racial slur appears in three song book titles. A term for intersex people that is now considered offensive also appears in a book title. Ultimately, I chose not to exclude offensive material from the datasets because that would compromise their accuracy as historical records and suppress evidence of ideas that were reflected and amplified through advertising material.17 However, users should be aware that they may encounter it. Instructors sharing the datasets with students may wish to flag this in advance.
Second, I needed to consider the impact of publishing identifying information. Tara Robertson has written about how digitization can expose living people to serious harm by making evidence of their past involvement in creating sexually explicit material far easier to find than they could have anticipated.18 As Colette Colligan has argued, digital tools that identify, or could facilitate the identification of, historical figures involved in making and selling sexual material may also lead to the 'unmasking' of figures who would not have wanted their involvement known.19 However, discovery was a risk of advertising that the people whose activities are documented in these datasets would have knowingly assumed: state authorities and anti-vice groups cited advertising material in obscenity trials and did not hide their use of this material in their own investigations.20 The youngest of these people also died more than a century ago. For these reasons, I’m comfortable with publishing the datasets.
Description
See the README for a description of the files.
Versioning
Please report any errors to sarah.bull@torontomu.ca. The datasets will be updated with any necessary amendments on GitHub at https://github.com/sarahebull/advertising_pornography_1822_to_1870 up to three times per year.
Licensing
This dataset package is published under a CC-BY 4.0 license.
The primary sources used to create the datasets in this package were all published before 1871 and are in the public domain. Organizations providing access to the sources were consulted and confirmed that I am free to publish content derived from them in a dataset.
Acknowledgements
I am grateful to Rebecca Baumann, Head of Curatorial Services and Curator of Modern Books at the Lilly Library for discussing this project with me over email and sharing her knowledge about the Michael Sadleir Collection of Ephemera. Thanks to Colette Colligan, Kathleen Lubey, and Lisa Z. Sigel for offering advice about how to classify some items.
Endnotes
Iain McCalman, Radical Underworld: Prophets, Revolutionaries and Pornographers in London, 1795-1840 (Cambridge University Press, 1988), 204-231.
See Lisa Z. Sigel, "Methodological Pitfalls in the History of Pornography" in Anna E. Clark and Elizabeth Williams, eds, Sources and Methods in the History of Sexuality (Routledge, 2025), 41-53 for a fuller discussion of challenges involved in studying these histories.
Peter Mendes, Clandestine Erotic Fiction in English 1800–1930: A Bibliographical Study (1993, repr. Routledge, 2016), 421.
See Sam Bailey, “Erotic Books in Eighteenth-Century England” (PhD thesis, Newcastle University, 2025), 213-275 for an excellent recent discussion of such collections’ histories.
See Colette Colligan, “Digital Discovery and Fake Imprints: Unmasking Turn-of-the-Century Pornographers in Paris.” Book History 22, no. 1 (2019): 249-279 for a fuller discussion.
See Mendes, Clandestine and Pisanus Fraxi [Henry Spencer Ashbee], Index Librorum Prohibitorum: Being Notes Bio- Biblio-Icono-graphical and Critical, on Curious and Uncommon Books (London: Privately Printed, 1877), Centuria Librorum Absconditorum (London: Privately Printed, 1879) and Catena Librorum Tacendorum (London: Privately Printed, 1885). Digitized scans of Ashbee’s bibliographies are presently available through the Hathi Trust Digital Library and Google Books.
As Meredith McGill shows, the boundary between the categories of “book” and “pamphlet” are blurry anyway, so I don’t feel like this is an awful decision. See McGill, “Books on the Loose” in Alexandra Gillespie and Deidre Lynch, eds, The Unfinished Book (Oxford University Press, 2020), 79-93, 79-80.
Unfortunately, these restrictions are quite common for collections of digitized historical periodicals. See Ruth Ahnert, Katherine McDonough, and Daniel C. S. Wilson. “Show Me the Data: New Practices for Historical Sources.” Transactions of the Royal Historical Society, First View (2026), 6. DOI: https://doi.org/10.1017/S0080440126100632
For example, if Edward Duncombe placed 100 copies of an advertisement with the same wording in 100 periodicals in 1850, I recorded one instance of its publication. If he published three advertisements with different wording or items listed for sale in the same periodical in 1850, I recorded one instance of each advertisement’s publication. Notably, William Dugdale issued some advertisements in radical periodicals that clearly aimed to attract readers who were primarily interested in radical material. I chose not to include these advertisements in the dataset.
Indeed, pornographers advertised in periodicals published outside London more frequently than this dataset suggests. See Sarah Bull, Selling Sexual Knowledge: Medical Publishing and Obscenity in Victorian Britain (Cambridge University Press, 2025), 47.
Bull, Selling Sexual Knowledge, 22.
For instance, compare the frontispiece in The Secrets of Nature Revealed; or, The Mysteries of Human Procreation and Copulation Considered and Explained (London: Printed for the Booksellers in Town and Country, 1832), Cup.365.a.36, British Library with descriptions of the frontispiece in the Sales Catalogues dataset.
Sigel, “Methodological Pitfalls,” 41.
Mendes, Clandestine Erotic Fiction, 144.
See Kathleen Lubey, What Pornography Knows: Sex and Social Protest since the Eighteenth Century (Stanford University Press, 2022) for rich investigation of this history of revision and its implications.
Catherine D’Ignazio and Lauren F. Klein, Data Feminism (MIT Press, 2023).’
See Lisa Z. Sigel, “Preserving ‘the intellectual garbage of the past’: A Call to Action.” Porn Studies 11, no. 1 (2024): 69-82, which discusses the dangers of obfuscating or erasing this kind of evidence.
Tara Robertson, “digitization: just because you can, doesn’t mean you should,” tara robertson, March 20, 2016, https://tararobertson.ca/2016/oob/.
Colligan, “Digital Discovery,” 273-275.
For examples, see “[C]ourt of the King’s Bench,” London Courier and Evening Gazette, November 19, 1829, 3 and “Frederick Hunt, alias John Brooks,” Sun (London), 17 July 1849, 8.