Fictional Systems:Mass-Digitization, Network Analysis, and Nineteenth-Century Australian Newspapers
Based on an analysis of the largest collection of mass-digitized newspapers available internationally, this article critiques current approaches to digital periodical studies, particularly relating to network analysis, while radically revising existing accounts of fiction reprinting and syndication in nineteenth-century Australia. It challenges the perceived dominance of Tillotson’s Fiction Bureau in this market and the associated ascendancy of syndicated British fiction over local writing. Turning to the critically neglected provincial press, it shows that these newspapers published and reprinted more fiction than their metropolitan counterparts. This material was supplied by an extensive, active, and hitherto essentially unrecognized array of syndication agencies operating within and beyond the colonies.
Among the extensive volume of fiction published in nineteenth-century newspapers, some titles appeared only once in a single publication for a single readership. But many were published multiple times in various forms and locations as part of a broad culture of reprinting and repurposing content.1 Fiction reprinting has been studied for the insights it enables into the social, commercial, and institutional operations and structures underpinning the production and circulation of periodical fiction during this period.2 However, the reliance of such studies on manually searching analogue archives has meant they are based on relatively small and selective samples of fiction in particular newspapers (typically major metropolitan titles), by particular authors (predominantly canonical figures), or as recorded in particular records of syndication agencies (those that have survived). Now, as with so many other areas of literary and book history, the possibilities for periodicals research have been transformed by the mass digitization of periodical content.3
This article radically revises existing accounts of fiction reprinting in nineteenth-century Australian newspapers based on a sample of over 9,200 “extended” fiction titles (mostly serial fiction as well as some very long stories completed in a single issue).4 This fiction appeared in over 250 newspapers and was identified through analysis of the largest mass-digitized collection of historical newspapers available internationally: the National Library of Australia’s Trove database.5 Existing accounts emphasise the dominance of Tillotson’s Fiction Bureau and the associated ascendancy of syndicated British fiction over local writing. I demonstrate that Tillotson’s was one participant among many in the colonial market, and I offer a new account of the nature, timing, and effect of its publishing activities. Previously, Tillotson’s was only associated with major city periodicals. I [End Page 100] show that it primarily engaged with second-tier metropolitan and provincial newspapers and that this occurred earlier, and more systematically, than has been recognised. The consistent presence of local writing for at least a decade after the arrival of syndicated British fiction refutes the claim that Tillotson’s and other overseas agencies ended opportunities for colonial authors. Moving beyond the practices of known agencies and agents, I confirm a significant shift in syndication practices in the 1890s, while demonstrating the role of specific Australian metropolitan newspapers in sourcing and distributing fiction for the colonies.
Turning to the provincial press—which has previously received almost no attention—my analysis of fiction reprinting reveals an entirely new set of activities and actors. Significantly, I show that provincial newspapers both published and reprinted more fiction than their metropolitan counterparts. Such reprinting involved a range of semi-formal editor- and author-led arrangements. But most extended fiction in provincial newspapers was supplied by an extensive, active, and hitherto unrecognised array of syndication agencies operating within and beyond the colonies. This new account of fiction reprinting and syndication in nineteenth-century Australian newspapers reveals a significantly more complex, varied, and populated array of processes and structures—local and global—than has been appreciated. It also highlights the extent to which past studies have approached the larger, previously largely intractable, newspaper archive through the lens of smaller, more tractable archives, and how this perspective has shaped and distorted understandings of colonial literary culture and its connection to the international fiction market.
Mass-Digitization and the Fictions of Network Analysis
Of course, digital resources and methods are far from neutral lenses: they impose their own partial view. Periodical studies has been at the forefront of humanities research in recognizing such partiality: contributions to the field were among the first to emphasise the large proportion of the archive that has not been digitized, as well as multiple issues affecting access to the contents of mass-digitized collections.6 However, periodical studies has yet to move coherently beyond acknowledging partiality to identifying its scope and effects and to devising strategies for interpreting results in that context. Indeed, a predominant framework through which the field imagines and increasingly represents engagement with the contents of mass-digitized collections—the network—tends to inhibit rather than enable nuanced historical analysis by obscuring the relationship between model and evidence. Elaborating these issues is an essential precondition for a study of fiction reprinting that relies upon data derived from a mass-digitized collection. The increasing importance of enormous, though inevitably [End Page 101] incomplete, datasets for periodical studies also makes these issues important for clarifying directions in the field more broadly.
As previously noted, Trove is the largest collection of digitized historical newspapers available internationally. On the date I ceased harvesting fiction for this article (July 16, 2015), Trove made 17,620,635 searchable pages available, compared with 9,728,249 pages for Chronicling America, 11,162,283 pages for the British Newspaper Archive, and 10 million pages for Europeana Newspapers.7 Although this page count is impressive, I estimate that Trove’s holdings represent approximately one-fifth of nineteenth-century Australian newspapers (based on a comparison of the Australian newspapers digitized and those indexed in advertising manuals of the period).8 For digital periodical research in general, this proportion should underscore the partiality of other major mass-digitized newspaper collections, where the number of pages available is significantly less than in Trove, even as the number of historical newspapers was considerably greater.9 Although most Australian newspapers are consequently omitted from this study of reprinting, I feel confident in describing my dataset as broadly representative, based on further comparison of my dataset with historical records. I make this assertion, however, with the caveats that South Australian and metropolitan newspapers, those from colonies with smaller populations, and those operating earlier in the century are somewhat overrepresented.10
I also believe that my paratextual method for analysing Trove has identified most of the extended fiction in newspapers digitized at the time I ceased harvesting data. Due to its interaction with the features of Trove, this method is largely unaffected by Optical Character Recognition errors.11 However, two issues—one arising from the method, the other from collection practices—did affect the type and range of fiction I discovered. The first issue was the necessary limit on the number of paratextual terms used to identify relevant results. “Chapter,” “serial,” “story,” “novelist,” “tales,” “sketches,” and “storyteller” were effective in returning instances of fiction in digitized newspapers. But additional terms would extend the discovery of fiction. Ultimately, I had to balance the number and utility of the terms I used against the finite time available to conduct this study (a condition of all research but one especially pertinent to the task of creating a stable, analysable dataset from an ever-expanding mass-digitized collection). Discovery of fiction in provincial newspapers was specifically impacted by the patchiness of analogue holdings for such publications and by another well-known problem for periodical studies: the routine exclusion of supplements (where most fiction was published in provincial newspapers) from the collection procedures underpinning Trove.12 These issues, combined with the general underrepresentation of provincial newspapers, mean that my findings understate the presence and reprinting of fiction outside metropolitan centres. [End Page 102]
Determining the viability of any dataset employed in periodical studies should be premised on a discussion of the scope the collection(s), the means of investigation, and the representativeness of the derived data. But the methods for representing and interpreting data also require careful consideration. Given my focus on newspapers that published the same fiction, network analysis would appear to be the obvious choice for this project. The reason for the method’s popularity is clear: its depiction of edges (relationships) between nodes (entities) mirrors an established, system-based understanding of print culture. Applied to the extensive datasets harvested from mass-digitized collections, the attractive network visualizations offered by programs such as Gephi appear to bring connections and configurations within periodical culture literally into view. However, at least as it is currently employed in periodical studies (maybe inevitably for research based on mass-digitized collections), network analysis inhibits the effective engagement with historical evidence. In particular, a focus on network visualization impedes scholars’ understanding of the evidence available to construct and interpret such models and creates perhaps insurmountable barriers to recognising and accommodating the evidence that is absent.
Perhaps encouraged by the routine designation of digital methods as “distant reading”—perhaps necessitated by a lack of statistical literacy—scholars in periodical studies tend to present and approach the results of network analysis in visual representations that can be interpreted or “read” to discover the operations of historical systems.13 The most basic way that this strategy impedes apprehension of historical evidence is by rendering implicit the decisions and assumptions by which literary data are constituted and arranged. Humanities researchers increasingly recognise data as artefacts rather than facts14 and algorithms as arguments that should not be black-boxed.15 But presenting network models as visual images—without publishing the data underpinning and produced by them—precludes assessment of these underlying procedures and their effects.
When the data underpinning the visualization are unavailable, the default position is to accept network connections as categorically meaningful and equivalent. This default risks mistaking the effects of data construction for historical processes. It is also increasingly likely as the scale of data increases and/or when such data are derived from automatic data mining, including the mining of mass-digitized collections. (Both situations limit the scholar’s capacity to confirm the nature of the individual entities and relationships represented.) For analyses of fiction reprinting, for instance, assuming the equivalence of connections created by newspapers publishing the same story conflates the multiple possible routes by which a title is obtained: some newspaper editors might have bought the story from the author, others from those purchasing newspapers or from a syndication [End Page 103] agency. Still others might have “borrowed” it from another newspaper without payment and with or without acknowledgement. While periodical research seeks to understand these underlying processes, the appearance of meaning and equivalence presented by connections in network models serves to deflect attention from the range, complexity, and conceivable contradiction of historical phenomena.
More generally, network visualizations compound the danger of anachronism associated with metaphoric references to the past in terms of networks. As a number of historians have argued, such metaphors risk projecting “contemporary, much faster, networked flows”—most obviously, those of the Internet—onto the historical context.16 Translating metaphor into material form increases the rhetorical impact of this projection, with the sense of immediacy, uniformity, and cohesion presented by network visualizations working against recognition of the specific and variable distances, extended temporalities, and complicated social, economic, and political negotiations involved in nineteenth-century periodical culture.
In the embrace of network analysis in periodical studies, these challenges have not been adequately articulated. But individual projects do employ various strategies to forestall such misapprehension of the available evidence. The Viral Texts Project publishes the data it harvests from Chronicling America and describes data construction. When focusing on specific instances of reprinting rather than the large networks arising from analysis of all titles, Ryan Cordell offers nuanced insights into the operations of early American print culture.17 Richard So and Hoyt Long distinguish between the edges in their network model and connections in Modernist literary culture by carefully delineating the assumptions underlying their data construction.18 Likewise, in her work on nineteenth-century genre formation, Anne DeWitt avoids mistaking the potential patterns arising from data mining for historical processes by reading each of the “thousands” of articles resulting from searching six databases for seven theological titles. This approach raises the problem of evidentiary excess that network analysis is intended to overcome, but it means that all 355 articles in her model meet her definition of genre formation (the claim by a reviewer of likeness between two or more titles).19 While all of these projects rely on visual representations of networks—and indeed, DeWitt identifies the value of network analysis in terms of the “advances in the visualization of data” it offers—such strategies avoid the key challenges this method presents to apprehending the available evidence.
They do not, however, counter the more pernicious problem of the representational approach to network analysis: its incapacity to identify and accommodate the effects of evidence that is unavailable for modeling. For a number of digital methods employed in periodical studies, it is sufficient to establish a broadly representative dataset (as I do above) or one where [End Page 104] areas of partiality are identified and taken into account in subsequent investigations. None of the above projects provides such an assessment.20 But even if they did, this would not constitute a sufficient basis for network analysis because the method dramatically amplifies the challenges of, and potential misconstructions arising from, working with partial data.
The degree of contingency in network models is poorly understood by humanities scholars and is inadequately addressed by an approach based on visualization. With the exception of geospatial formats, network models arrange nodes entirely according to the proprieties of the available data. In a force-directed graph, for instance, algorithms position nodes based on the number and strength of edges they share with other nodes. As a result, adding new nodes or edges—an inescapable prospect in a field where only a very small portion of the archive is digitized—will always change the position of all entities depicted, often radically. The considerable gaps in what is available to be modelled in periodical studies mean that network visualizations based on mining mass-digitized collections invariably present fictitious systems: arrangements that are a function of what has been digitized as much if not more so than how a literary-historical system operated. Projects that base historical arguments on the structure of network models (for instance, describing the “betweenness” of a particular work, author, or site of publication) ignore this radical contingency and implicitly maintain that all data (or all data conceivably relevant to understanding a particular historical system) is available.21 This approach reinforces the false sense of completeness—of coherent and self-contained systems—projected by visual models of networks.
The apparent completeness of network models obscures another gap in the evidence needed to interpret such structures: the absence of the documents explaining the nature and function of the entities and relationships proposed. Although mass-digitization is understood in terms of evidentiary excess, it concurrently creates a profound evidentiary imbalance for digital periodical studies, between extensive (though incomplete) information on the contents of periodicals and the limited availability of the documents needed to understand the actors and institutions responsible for creating and distributing those contents. In this project, for example, the evidence needed to determine the basis of different instances of reprinting is often unavailable: in particular, for most syndication agencies, records no longer exist. Indeed, a key reason Tillotson’s has received so much attention in studies of fiction reprinting is that its archive, though “scrappy” with multiple gaps, is a comparatively rich resource in a context where many of the names of syndication agencies, let alone their business activities, have been lost to history.22
Statistical analysis provides an alternative to a representational approach to network analysis, one capable of identifying and accommodating [End Page 105] incompleteness in the data available to be modelled and interpreted. The measures I am referring to are not those built into programs such as Gephi (for instance, graph density, modularity, or weighted degree). These characterise the effects of network modelling on the available dataset: they do not accommodate gaps in evidence. Scientific and social scientific applications of network analysis employ alternative statistical approaches to this end. Measures of probability, for instance, assess the likelihood that the stated characteristics of a modelled network would remain true if all data were available, while “forest” networks address questions of causality when the processes underpinning particular relationships are unknown but are from a finite set.23 Such measures recognise that questions relating to system structures and their dynamics are especially sensitive to data completeness. Even when analysis is based on a representative dataset, for network modelling to serve as a justifiable foundation for argument the researcher must establish the low likelihood that results were a sampling effect of the available data or produced by random chance.
Certainly, it would be possible to apply statistical approaches to characterising and accommodating gaps in data generated from mass-digitized collections, including the data used in this project.24 Some digital humanities scholars advocate precisely this move to more sophisticated statistical methods.25 But even if periodical scholars developed the literacies needed to conduct and interpret such measures—and narrowed the form of the questions asked of network models accordingly to the structural effects of interrupting relationships rather than the workings of the system—I do not think the results of network analysis would be adequate as historical evidence. The probability measures needed to model systems based on the highly incomplete information provided by mass-digitized periodical collections are at odds with the desire for documentary evidence that is so central to historical argument. Literary historians focus on what occurred and why—not what might have taken place based on a series of assumptions and probabilities. And even if mass-digitization continues to the extent that probability measures relating to periodical contents could be employed without too many statistical accommodations (a situation that appears a long way in the future if achievable at all), the inevitable distinction between such content and the evidence needed to understand underlying historical processes means that network analysis could only offer part of the methodological toolkit for any study.26
In light of these issues, I have not based any of the arguments below on the outcomes of network analysis, nor do I offer any network visualizations. But I do employ network analysis for specific practical and exploratory purposes. Gephi’s “multimodal networks projection” feature enabled me to create a more manageable dataset by converting thousands of connections [End Page 106] between fictional titles and newspapers into hundreds of connections between newspapers based on the titles they published in common. The network models I produced from this dataset showed interesting patterns: for instance, that certain newspapers (such as Melbourne’s Leader) were highly connected and that metropolitan and provincial newspapers tended to cluster together with few connections between them. However, in exploring these models I remained acutely conscious of their contingency and partiality as algorithmic projections of an available dataset that excludes most of the actors and enterprises, local and global, implicated in the system I am investigating (not only the four-fifths of nineteenth-century Australian newspapers absent from Trove but other colonial and overseas periodicals, authors, syndication agencies, literary agents, publishers, and so on involved with fiction reprinting in the colonies).
In other words, I treated the connections and patterns proposed by network analysis as potential indicators, rather than evidence, of reprinting practices. To establish the basis for a historical argument, I approached these results with the type of questions that scholars have long asked in periodical studies, and I based my answers on forms of evidence the field has traditionally relied upon. If newspapers published multiple titles in common, I asked: Who owned these newspapers? What was the physical distance between them? What was the specific sequencing of republication for different titles and did it remain the same over time? Next, I studied digitized newspaper pages to query: Are page layout and typographical features the same in all instances of publication? Are illustrations—and the same illustrations—present? Is the source of the fiction acknowledged? Finally, I searched critical bibliographies and published records of syndication agencies to find the answers to other questions such as: Who else published this story? How much were authors paid, who represented them, and what other authors did they work with?
The resulting perspective on nineteenth-century reprinting practices is significantly extended by mass-digitization and by the exploratory capacities of network analysis. But it is not based exclusively on the contents of a mass-digitized collection and the alignments suggested by a digital method for analysing them, nor is it a comprehensive view. Rather than delineating a literary historical system, my analysis constructs a narrative out of multiple pieces of evidence that enable important insights but also produce multiple dead ends: places where gaps in the evidence (relating to periodical content and/or to the documents needed to interpret it) limit what I can know. [End Page 107]
In existing accounts of the reprinting of fiction in nineteenth-century Australian newspapers, the mid- to late 1880s are identified as a period of dramatic change marked by the arrival and immediate dominance of British syndication agencies, principally Tillotson’s. Critics agree that prior to this time no established systems existed for sourcing overseas content. According to Toni Johnson-Woods, it is a “mystery” how imported serials came to Australia during this period, but it is likely that piracy was involved, especially where American fiction was concerned.27 Others have described how colonial newspaper editors obtained fiction through contracts with individual British authors28 and through “unauthorized ‘borrowings,’” with short fiction more likely coming from local publications and serial fiction coming from overseas.29
Graham Law ascribes Tillotson’s dedicated involvement with the colonial market to “financial pressures in their home market.”30 Whereas the company experienced strong growth in sales to English newspapers from its beginnings in 1873 to the mid-1880s, towards the end of that decade Tillotson’s was compelled “to search more energetically for returns elsewhere … [through] ventures into America, the Colonies, and Europe.”31 In making this move, Law argues, the agency dealt only with “major city journals.”32 The “standard arrangement for works by well-known writers like [Mary Elizabeth] Braddon” was for Tillotson’s “to offer serial rights in a single colony for £75, or entire Australian and New Zealand rights for £100, thus leaving a Colonial editor or agent to sell on copy to other journals.”33 Paul Eggert concurs with Law’s timing when he argues that overseas “agents … saturated the local market with imported serials” from the mid-1880s,34 while others join Law in emphasising the particular dominance of Tillotson’s. Johnson-Woods, for instance, notes that Tillotson’s provided “nearly all of [the] imported stories” in major metropolitan newspapers.35 Scholars also generally agree that the entry of overseas syndicates into the Australian market had a deleterious effect on local literary production. Christopher Hilliard argues that fiction was supplied to the Australian colonies so cheaply by overseas syndication agencies, Tillotson’s in particular, that local literary production was significantly constrained.36
This established account would lead us to anticipate relatively haphazard and minor incidents of fiction reprinting in nineteenth-century Australian newspapers until the mid- to late 1880s, followed by a sudden and substantial increase and consistency in the practice. The solid grey line in figure 1, indicating the proportion of titles reprinted among metropolitan publications per year, shows the opposite of this trend.37 High (though uneven) rates of reprinting prior to the mid-1880s were followed by an overall decline. These results require qualification, however, due to a phenomenon [End Page 108] I will call “companion reprinting.” From the late 1850s, multiple daily metropolitan newspapers established weekly companions. As might be expected—and as the dotted black line in figure 1 indicates—these jointly owned, and often jointly edited, newspapers frequently published the same stories.
Most fiction reprinted between metropolitan newspapers prior to the mid-1880s falls into this category of companion reprinting. The daily Brisbane Courier and the weekly Queenslander were the first to engage in the practice routinely, with a particular emphasis on American fiction. (Perhaps the editors thought the content of these stories would speak to Queensland’s frontier society; more likely, they felt justified in publishing such fiction for free due to the lack of American acknowledgement of international copyright.)38 A number of other daily and weekly companions (including the Evening Journal and Adelaide Observer in South Australia, the Telegraph and Week in Queensland, and the Evening News and Australian Town and Country Journal in New South Wales) also frequently published the same stories. Still others—among them the largest and most culturally significant metropolitan newspapers—published a significant number of stories individually, but rarely, if ever, together.39
When companion reprinting is excluded, rates of fiction reprinting among metropolitan newspapers more closely resemble the established narrative. As the solid black line in figure 1 indicates, the proportion of reprinted titles increased across the nineteenth century, from under 10 percent prior to the mid-1880s to between 10 and 20 percent in the following decade, albeit with a sharp decline in the second half of the 1890s. This [End Page 109] period of more extensive reprinting corresponds with the time Tillotson’s supposedly entered and dominated the colonial market, but it demonstrates nothing of the dramatic and abrupt shift in fiction reprinting that might be expected. Comparing titles syndicated by Tillotson’s with fiction identified in this study further disrupts the prevailing account of that company’s activities.40 From 1880, almost all fiction syndicated by Tillotson’s appeared either in that same or in the following year in one or more colonial newspapers.
Such systematic involvement with colonial newspapers suggests that Tillotson’s acted offensively rather than defensively in its international expansion, while the alignment between the authors published in colonial newspapers prior to 1880 and those syndicated by Tillotson’s after this time suggests an explanation for this earlier and alternative mode of engagement. Well before Tillotson’s was created, multiple authors later associated with that company were published extensively in colonial newspapers. In addition to works by Mary Elizabeth Braddon, in the decade prior to 1875 multiple titles by Wilkie Collins, B. L. Farjeon, George Manville Fenn, James Payn, Charles Reade, and F. W. Robinson appeared in metropolitan newspapers, and many were reprinted two or more times. The second half of the 1870s witnessed more fiction by these and other authors later syndicated by Tillotson’s, including work by Walter Besant, William Black, Eliza Lynn Linton, Justin McCarthy, George Macdonald, Margaret Oliphant, and Dora Russell.
Pirating most likely explains early, though much less extensive, appearances by some of these same high-profile British authors in provincial Australian newspapers. But metropolitan publications frequently published well-known British authors with explicit statements about copyright. Some of these notes were general—for instance, many indicated that a story was “published by special arrangement with the author” while others noted that the “right of republishing” was purchased by the proprietors of particular newspapers. Other notices were highly specific regarding the extent and nature of copyright—for instance, claiming exclusive rights to reissuing the story in a particular colony. Combined with what Sarah Ailwood and Maree Sainsbury describe as the exceptional adherence of the Australian colonies, of all British dominions, to imperial copyright law, such prominent and prevalent assertions of copyright strongly imply that these well-known British authors were published in metropolitan newspapers under paid contract.41 By 1870, then, many of the very authors Tillotson’s would later seek to court were already negotiating extensively with the Australian press, in person or through agents, a practice continued throughout the decade. Instead of waiting until the mid- to late 1880s, when there was a decline in profits from syndication in Britain, it seems [End Page 110] much more likely that authors urged Tillotson’s to engage with the established Australian market from the outset.
The type of newspapers Tillotson’s dealt with also reconfigures existing perceptions of its relationship to the colonial market and explains why its earlier, systematic involvement has been overlooked. Whereas previous studies have stated or assumed that Tillotson’s worked only with major metropolitan publications and targeted their analyses accordingly, in fact the company was much more likely to engage with second-tier metropolitan newspapers. The South Australian Chronicle was a leading colonial customer of Tillotson’s, as were the Adelaide Observer and Evening News from South Australia and the Week and Telegraph from Queensland. In the 1880s, and especially the 1890s, Tillotson’s was also increasingly likely to contract with provincial publications, at first targeting the earlier and larger newspapers in this category, such as the Bendigo Advertiser, Capricornian, Goulburn Herald, and Morning Bulletin, and then proceeding to multiple smaller enterprises, including the Barrier Miner, Clarence and Richmond Examiner, Elsternwick Leader, Launceston Daily Telegraph, Launceston Examiner, and Oakleigh Leader. By comparison, the major metropolitan dailies and weeklies typically associated with Tillotson’s—the Age, Australian Town and Country Journal, Illustrated Sydney News, Leader, and Sydney Mail—published relatively few titles syndicated by that company.
An exception to this latter trend occurs with what Law calls the “expensive serials” of the 1890s.42 Tillotson’s paid large amounts for these titles by prominent authors, which were not published in its own Bolton newspaper group. Law argues that this fiction was “purchased particularly or exclusively for the American market,”43 but this analysis shows that it was also acquired by major metropolitan newspapers, including the Age, Australian Town and Country Journal, Leader, and Sydney Mail.44 Although Tillotson’s engaged with these major periodicals, its primary involvement with second-tier metropolitan and provincial newspapers shows that it moved into the Australian market via the same approach it pursued in Britain: by sourcing fiction for newspapers that lacked the resources to pursue content independently. The focus of earlier studies on major metropolitan newspapers provides the obvious, practical reason why this parallel in Tillotson’s activities in Britain and Australia has been overlooked. But arguably, the notorious Australian “cultural cringe” has also played a role, encouraging the perception that Tillotson’s—as a British company, despite its provincial position in the home market—would naturally occupy a privileged position in the colonial cultural sphere, dealing only with the most prestigious newspapers.
We might perceive this same bias in the widespread view that Tillotson’s entry into Australia immediately ended opportunities for local authors. [End Page 111] The assumption that Australian newspaper editors would invariably select the imported over the local product is challenged by the results in figure 2, which indicate the proportion of American, Australian, British, and “other” fiction published. The solid lines in this graph indicate the yearly proportions overall (including titles reprinted multiple times in a single year), while the dotted lines represent proportions of unique titles (which is a more accurate means of assessing opportunities for local authors, who were less likely than British writers to have their fiction reprinted in metropolitan newspapers). Although British writing is clearly dominant, Australian fiction has a sustained presence throughout the 1870s and much of the 1880s, often comprising over 30 percent of the fiction available and, as late as 1887, 28 percent of all known fiction and 33 percent of unique titles. In other words, for a decade after Tillotson’s entered the Australian market in the late 1870s and for seven years after the company began systematically to sell fiction to colonial newspapers, Australian authors clearly had opportunities for metropolitan publication.
The trends discussed thus far clearly indicate that Tillotson’s was not the predominant actor in colonial fiction publishing, as previous histories have claimed. But understanding what companies, individuals, and practices supplied fiction to colonial newspapers in the absence of this prior explanatory framework presents a challenge. Based predominantly on indexes of major metropolitan newspapers, advertisements in industry publications, and/or surviving correspondence, Graham Law, Charles Johanningsmeier, and others have noted the involvement of various overseas agents and agencies in the Australian market, including the major American enterprises McClure’s and Bacheller’s and the British literary agent A. P. Watt.45 Comparing the titles Watt syndicated in Britain with their appearance in Australian newspapers indicates that his role in colonial fiction publication was much more organised and consistent than has been hitherto understood.46 It is also clear, based on copyright descriptions in metropolitan newspapers, that Cassell’s was active in supplying the Australian newspaper market. However, without more information about the specific titles syndicated by these and other agents and agencies or knowledge of the terms under which they were contracted, it is impossible to be precise about the extent of their activities individually or in relation to Tillotson’s.
It is possible to gain a more general perspective on the syndication industry and its operations by assessing the presence of fiction in metropolitan newspapers written by approximately 100 authors associated with well-known British and American syndication agencies and agents.47 Comparing the contributions of these 100 or so authors with the rest of the field affirms the established and efficient mechanisms through which syndication agencies operated, in that the average number of titles by associated [End Page 112] authors in metropolitan newspapers is considerably higher than for other writers. Indeed, all but two of the top twenty most published authors in colonial metropolitan newspapers, and many of the top forty authors, were aligned with these particular organisations.48
Yet the perceived dominance of these particular agents and agencies in the Australian market is simultaneously challenged by the relatively small and declining contribution that associated authors make to fiction in colonial metropolitan newspapers. Figures 3 and 4 compare the proportion of fiction supplied by associated and non-associated authors: the former charting all writers, the latter focusing exclusively on British authors.49 As figure 3 shows, metropolitan newspapers published a high proportion of fiction by non-associated authors: with the exception of two years (1891 and 1892), this number was always over 60 percent and typically in excess of 70 percent. This result is especially surprising given the high average number of titles that associated authors contributed. Although my list of authors is undoubtedly incomplete and is small relative to the size of the overall publishing context, this result emphasises how much we do not know about the source of fiction in colonial newspapers. What Johnson-Woods describes as a “mystery” before 1870 remains largely a mystery after 1880. Certainly, figure 3 indicates a situation very distinct from Johnson-Woods’s claim that Tillotson’s supplied “nearly all” of the fiction imported into the colonies.
To some degree, figure 4 suggests a more recognisable narrative. It shows that the 100 or so authors associated with known syndication agencies and literary agents supplied the majority of British fiction in metro [End Page 113] politan colonial newspapers, including 70 percent of this fiction published between 1882 and 1892. However, the subsequent fall in this proportion, to 50 percent or less, suggests a significant shift in the supply of fiction to the colonies. Though less obvious, the same trend is present in figure 3, where the overall proportion of fiction by authors associated with these specific agencies falls from 40 percent in 1892 to 20 percent by the end of that decade.
[End Page 114]
Further reinforcing the sense of change in the early 1890s is the resonance between these results and two trends from previous graphs. The first is the decline in the proportion of fiction reprinted among metropolitan newspapers in the second half of the 1890s. As figure 1 shows, where 22 percent of extended fiction in metropolitan newspapers appeared in two or more (non-companion) periodicals in 1895, by 1899 this total fell to only 8 percent. The second is the decline in Australian fiction in these periodicals. Although not as definitive as would be expected from existing accounts of overseas agencies saturating the local market, the reduction in Australian writing shown in figure 2—from 28 percent of fiction (or 33 percent of unique titles) in 1887 to 21 percent (or 24 percent of unique titles) by 1899—implies a shift in the source of fiction for metropolitan colonial newspapers. The most probable explanation for these combined trends is increased competition in the Australian market from new overseas agencies. An increase in the number of syndication agencies offering overseas fiction at reduced prices would logically produce a decline in the market share of earlier syndicators, while increasing the presence of non-Australian fiction.50 Lower prices would also reduce the need for metropolitan newspapers to join together to purchase particular stories, thus explaining the reduced incidence of reprinting among such publications.
In fact, Elizabeth Morrison has already described the entry of new overseas syndicators and the growth in competition as features of the colonial fiction market in the 1890s.51 But her claim that these new companies were American is countered by the national origins of the fiction published. As figure 2 shows, British fiction increased as a proportion of titles in colonial metropolitan newspapers from the late 1880s, while American fiction remained stable and even declined. Although American companies are known to have syndicated British fiction, including in Australian newspapers, one would expect some growth in the presence of American titles if such companies constituted the majority of the competition in the market.52 The fact that British authors were responsible for about 60 to 70 percent of the fiction in colonial metropolitan newspapers in the 1890s suggests that some, if not most, of this increased competition was supplied by British enterprises.
While much about overseas influences on the Australian market in the late 1880s and 1890s remains unclear, the manner in which particular newspapers engaged in fiction reprinting before and during this time offers new insights into the local industry’s operations and structure. The Leader and the South Australian Chronicle, along with three pairs of companion publications—the Brisbane Courier and Queenslander, the Evening Journal and Adelaide Observer, and in the 1890s, the Telegraph and Week—emerge as being so central to the colonial metropolitan culture of reprinting [End Page 115] that, in the available dataset, few instances of the phenomenon do not involve one or more of these newspapers. The fact that these papers also published the greatest volume of fiction overall affirms the importance of reprinting as the means by which metropolitan newspapers accessed content.53 More specifically, the sequence of reprintings among these papers and others indicates the important roles they played in distributing fiction throughout the colonies.
Table 1 summarises instances of reprinting involving these newspapers and indicates whether they published these materials first or secondarily. It shows that the Leader routinely published fiction that subsequently appeared in other Australian newspapers, suggesting that its editors sourced and sold fiction within the colonies, particularly to those newspapers I have described as second-tier metropolitan publications. Most of the reprinted fiction initially published in the Leader was written by authors associated with known syndication agents and agencies. But only a small number of these titles can be tied directly to Tillotson’s. In this respect, the Leader’s position in the colonial culture of reprinting demonstrates the practice Law proposed as standard—for Tillotson’s to sell fiction by well-known authors to a single metropolitan publication, leaving it to distribute rights within the colonies—while emphasizing that Tillotson’s was not the only company pursuing this approach.54 The Leader sourced its fiction from international syndication networks rather than a single company.
Although involved in almost as many incidents of reprinting as the Leader, the South Australian Chronicle, until the 1890s, tended to adopt the opposite approach—featuring fiction already published in other colonial newspapers. Although the Leader was its single main source of fiction, the South Australian Chronicle reprinted titles from a range of other newspapers, including major metropolitan publications (such as the Age, Illustrated Sydney News, Australian Town and Country Journal, and Australasian) and smaller metropolitan newspapers (such as the Express and Telegraph, Queenslander, Telegraph, Week, and West Australian). This wide range of sources refutes the view that only major metropolitan periodicals supplied fiction to other colonial publications, while further dismantling existing perceptions of Tillotson’s dominance in the market. Earlier, I identified the South Australian Chronicle as one of Tillotson’s main colonial customers, alone or in conjunction with other periodicals. Here, Tillotson’s is repositioned as only one source of its fiction among other enterprises, including numerous colonial newspapers.
The three pairs of companion newspapers listed in table 1 typically republished fiction by high-profile British authors that had originally been sourced from the Leader. In contrast, the fiction these papers published first and then supplied to other newspapers (including each other) was [End Page 116] by lesser-known (or unknown) authors. This latter sequence suggests that these papers, including the Brisbane Courier and Queenslander, the Adelaide Observer and Evening Journal, and in the 1890s, the Telegraph and Week, were colonial conduits for cheaper sources of fiction. In this respect, growth over time in the number and proportion of titles first published by these newspapers corresponds with a structural shift in the sources of colonial fiction in the late 1880s and 1890s. It suggests that these newspapers were contracting with newer fiction syndicators, providing a key avenue through which these enterprises entered the colonial market to compete with, and ultimately to substantially displace, established agencies.
The account offered thus far radically expands previous conceptions of colonial fiction reprinting and syndication, indicating a much more populated and dynamic system than has hitherto been recognised. The complexity of that system increases considerably when provincial newspapers, regarded by most existing scholarship as uninvolved in fiction publishing, are taken into account. The multiple semi-formal and formal systems of fiction distribution discovered in this context—operating for the most part, entirely apart from the metropolitan press—indicate new features of the circulation of fiction in the nineteenth century, both globally and within the Australian colonies. [End Page 117]
Figure 5 shows the number of titles published in, and the proportion reprinted among, provincial newspapers per year from 1865 to 1899.55 Prior to the mid-1870s, provincial newspapers published relatively little extended fiction, and the small amount of reprinting that occurred was from metropolitan (predominantly British but also Australian) periodicals. When fiction reprinting among provincial newspapers became more common, some of this activity occurred between companion publications, as was the case in a metropolitan context. However, in the provincial press the same trend occurred a decade later and involved only one pair of newspapers—the daily Rockhampton Morning Bulletin and weekly Capricornian—which in fact published more titles in common than any other newspapers in my sample.56
Even without these companion publications, the solid black line in figure 5 shows a clear correlation from the mid-1870s between the growth in extended fiction in provincial newspapers and incidences of reprinting among such publications. Indeed, almost as soon as reprinting began, it became a major—in some periods, the dominant—source of fiction for provincial newspapers, regularly comprising about 40 to 50 percent, and up to 60 percent, of titles published. In contrast, among non-companion metropolitan newspapers this figure only once exceeds 20 percent and is often less than 10 percent. Just as importantly, comparing figure 5 with figure 1 shows that provincial newspapers published dramatically more fiction than metropolitan papers. Even though fiction only became a common feature of provincial newspapers from the mid-1870s, this project has discovered approximately 5,200 works of fiction in their pages, compared with 3,800 in the pages of their metropolitan counterparts. Both findings are hugely important in and of themselves and bear repeating: fiction publication was more active and interrelated in the provincial than in the metropolitan press. This contradicts the orientation of existing scholarship, which has focused almost exclusively on metropolitan publications in its discussion of the structural and other features of reprinting.
Some reprinting among provincial newspapers resulted from editor- and author-led endeavours of varying formality. In addition to their shared publications, the Morning Bulletin and Capricornian published multiple stories in conjunction with other provincial newspapers, including the Armidale Express, Bendigo Advertiser, Clarence and Richmond Examiner, and South Bourke and Mornington Journal. These titles were typically acknowledged as having been reprinted from British periodicals (most often Chambers’s Edinburgh Journal), and the majority appeared first, by a few weeks, in the Morning Bulletin and Capricornian, although the reverse also occurred. This pattern of reprinting suggests an exchange system where the Morning Bulletin and/or Capricornian were sent to other [End Page 118] provincial editors in return for their newspapers. A more formal—though more limited—reprinting arrangement was practiced by the Goulburn Herald, at different times with the Hay Standard and Cootamundra Herald. The layout and timing of these publications indicates that the Goulburn Herald sold partly printed sheets to the other two newspapers, while the unattributed nature of these stories—even those by famous authors such as Wilkie Collins—suggests that the Goulburn Herald did not reduce this income stream by paying writers or intermediaries for the right to publish and reprint.57
Another semi-formal system of reprinting is associated with an enterprising local author, David Hennessey. As a journalist, editor, and publisher, Hennessey had access to the networks required to syndicate his fiction, and he did so with at least five different titles.58 There is also substantial evidence that his ambitions extended beyond the placement of his own work and led him to establish a number of publishing enterprises. One of these, Hennessey and Harper, advertised itself as “Authors’ Agents, Press Correspondents, Advertisement Contractors, Publishers, etc. etc.” with services including “Printing and Publishing of Books, Serial Stories, etc., arranged for in England or the Colonies.”59 Despite this claim, I have only discovered examples of Hennessey’s syndication of his own writing.60 Even so, his success in placing fiction in the provincial press represents a significantly more substantial instance of authorial syndication than the only other previously identified colonial example: James “Skipp” Borlase’s abortive attempt to establish a fiction syndication agency in the 1860s.61 [End Page 119]
While many editor- and author-led endeavours contributed to the practice of fiction reprinting, the vast majority of provincial newspapers employed formal syndicates. In contrast to the metropolitan context, where instances of reprinting typically involved two or three periodicals, provincial syndicates were extensive, encompassing multiple newspapers and titles. Identifying these syndicates is necessarily a provisional exercise, particularly during the 1890s, when stereotype and reprint columns, rather than ready-printed and largely identical supplements, became increasingly common. However, patterns of reprinting in the available sample indicate at least eleven substantial syndicates operating in the provincial market, which are summarised in table 2.62 Given the large number of nineteenth-century Australian newspapers (particularly provincial ones) that have not been digitized and the tendency for collection practices to exclude supplements, I have no doubt that the number of newspapers involved in these syndicates—and probably the number of syndicates in operation—was substantially greater than I have been able to discern. But even on the available evidence, with newspapers being the main source of fiction in the Australian colonies and more fiction appearing in provincial than metropolitan publications, these provincial syndicates should be recognised as the leading publishers of fiction for nineteenth-century Australian readers.63
To my knowledge, the first syndicate in table 2 is the only one that has been identified and studied. Morrison, one of the few scholars to consider provincial colonial newspapers in any detail, identifies this syndicate as having been owned and managed by Donald Cameron under the Cameron, Laing and Co. imprint.64 To Morrison’s excellent account I can add only a little. Although many early and later titles in the syndicate were from overseas (especially America), Cameron, Laing and Co. focused on local fiction. Morrison emphasises the significance of this investment in colonial writing. However, I believe her claim could be pushed further by identifying this local syndicate as one of the most prolific publishers of Australian novels, at least in the nineteenth century and probably well into the twentieth century.65 Given that newspapers were one of the few avenues of publication for colonial authors, only a small number of the titles syndicated by Cameron, Laing and Co. were issued as books. As a consequence, many are missing from the existing Australian bibliographical record, even though they were written by well-known and popular writers of the period. Comparing the sequence of titles published by this syndicate with those in New Zealand newspapers (digitized through Papers Past) suggests that Cameron, Laing and Co. operated beyond the Australian colonies as well as between them.66 Whereas Morrison proposes that the syndicate ended in 1888,67 the evidence suggests that it continued beyond that time until at least 1892.68 [End Page 120]
[End Page 122]
While I do not know who owned the other enterprises listed in table 2, their practices, including the types of fiction they issued, help to characterise the different syndicates and the provincial fiction market in various ways. The most notable dynamic is a rupture in the early 1890s, when syndicates 1 through 5 ceased operating and syndicates 6 through 9 began. The syndicates in the first group had the same basic format: two partly printed sheets, usually published as a supplement to the newspaper (an additional two or four pages). Supplements typically started with a poem, followed by an instalment of a story and sometimes a short story or two. The remainder was comprised of what Morrison describes, in her study of Cameron, Laing and Co., as a “melange of reprinted material, most of the latter extracted from overseas—chiefly American—magazines and newspapers.”69 Yet within this standard format, the syndicates demonstrated substantial variation.
Based on the available evidence, the syndicates differed markedly in scale, with syndicates 3 and 5 noticeably smaller than the others. Whereas most offered a mixture of short, medium-length, and full-length serial fiction, with a preponderance of the latter, syndicate 2 mainly dealt in short serials that were completed in two or three issues. The short stories serialised in syndicate 2 were predominantly of overseas origin, with the large proportion of unknown authors suggesting unauthorised borrowings from overseas periodicals. But the Australian content incorporated elsewhere in its supplements—in the form of poems and illustrations—implies that syndicate 2, like syndicate 1, was of local origin.
Syndicate 4 published no local fiction and was exceptional in other ways, too. Whereas the other syndicates in this period published one lengthy serial at a time, syndicate 4 offered multiple serials concurrently.70 In addition, whereas other syndicates published the same sequence of titles, often months apart in different newspapers, in syndicate 4 publication occurred within a few days across all periodicals.71 This clear difference in timing suggests that syndicate 4 was highly organised from its origins, whereas the other syndicates grew more organically, with newspapers able to join at different stages, receiving the full run of partly printed sheets in sequence.
The selection of authors published by syndicate 4 indicates that it was closely aligned with the international fiction market of the period. Many of its titles were by high-profile British writers associated with known syndication agents and agencies (including titles by Walter Besant, Arthur Conan Doyle, George Manville Fenn, G. A. Henty, Arthur Quiller-Couch, W. Clark Russell, and Robert Louis Stevenson), and two of its stories—Mary Elizabeth Braddon’s “Like and Unlike” and Hall Caine’s “The Bondman”—were specifically syndicated by Tillotson’s. (In fact, they appeared in provincial newspapers before colonial metropolitan publications.) Whether organised from within the colonies or imported, syndicate [End Page 123] 4 quickly won the market share with its distinct practices and well-known authors. Many newspapers transferred to this syndicate from other enterprises, especially Cameron, Laing, and Co. Yet even with this apparent success, syndicate 4 shared the fate of the other enterprises in this first group, ceasing operations in the early 1890s. At this stage, a new group entered the market, either out-competing earlier syndicates or filling a void left by their demise. Two further syndicates, 10 and 11, began operations in the final years of the nineteenth century.
A number of other features besides timing differentiate this second group of syndicates from the first. Whereas those in the first group traded in partly printed sheets, most in the second offered more flexible reprinting formats, allowing editors to incorporate syndicated contents (for instance, three columns’ worth for an instalment of the serial story) with their own advertising.72 Earlier syndicates can be clearly differentiated from each other, but this is less true of later enterprises. Although syndicates 6 and 10 featured local fiction quite regularly, the others published a more general, international mix of titles, including a substantial number by authors of unknown origins. Movement of newspapers between syndicates also occurred more regularly, suggesting greater competition in the market and the increased agency of provincial editors.
Of this second group, syndicates 6 and 10 were probably local. In addition to featuring local fiction, syndicate 6 included a small amount of local advertising on some of its partly printed pages, while syndicate 7 incorporated local content among its general-interest materials (for example, an article on the “Improvement of New South Wales Stock” in a syndicate largely comprised of provincial New South Wales newspapers). Syndicates 8 and 9 were probably American imports: both, but particularly 9, featured American fiction, while 8 incorporated advertising for American products and services (for instance, “Genuine Magic Soap,” “Patents” lawyers, and “Murray and Lanman’s Florida Water”). If American, they could be any of the multiple enterprises Johanningsmeier identifies as emerging in the 1890s, but for which little, if any, evidence survives.73 Based on the contents of the remaining two syndicates, 7 and 11 could be either local or American.74
These provincial syndicates present exciting possibilities for future research. Confirming that they were local enterprises would expand the history of Australian fiction publishing and displace the longstanding view that this activity was rare in the nineteenth century.75 Associating these syndicates with specific American or other overseas companies would add an important new transnational dimension to colonial periodical studies and nineteenth-century literary culture broadly. While I hope others might find evidence to support their own arguments in the sequence of titles I [End Page 124] have constructed, here we reach the limits of what the extensive sample of fiction used in this study can indicate. As discussed earlier, the mass-digitization of historical newspapers significantly expands access to periodical contents and offers an important new foundation for research. But this evidence is neither complete nor sufficient in and of itself. The sample I have employed, though representative, is only a partial reflection of the fiction found in nineteenth-century Australian newspapers, fiction that appeared as a consequence of institutional and social configurations and practices that are not often discernible from periodical contents and are only discoverable—if at all—based on other sources of evidence.
Even with many questions remaining, this study profoundly refigures existing conceptions of fiction reprinting in nineteenth-century Australian newspapers. Writing in the Melbourne Review in 1878, James Smith described Australian literature as having been eclipsed beneath the “shadow of England’s mighty and ever-spreading literature.”76 While this description resonates with claims made by subsequent literary historians and periodical scholars, trends in fiction reprinting in Australian newspapers indicate a considerably more complicated situation. At the time Smith was writing, and for at least a decade after, local fiction had a sustained presence in the pages of both metropolitan and provincial newspapers. In the former group, Tillotson’s Fiction Bureau was not the central and dominant influence, as has been proposed, but was one participant among many in a colonial market where both local and overseas enterprises played active roles.
Nor were metropolitan publications the dominant purveyors of fiction in the colonies: provincial newspapers published much more of this material, which was supplied by an extensive group of syndication agencies operating in the colonies and beyond. International fiction became more prevalent—with British fiction increasingly prominent in metropolitan newspapers and American fiction in provincial ones—at a time of decline in the importance of reprinting as a mechanism for attaining and distributing fiction in the colonies and, indeed, of newspapers as fiction publishers. Continuing mass-digitization will present further opportunities to discover and investigate fiction in nineteenth-century Australian newspapers and the means by which it was published and republished. But the view that British fiction and syndication agencies—let alone a single British company, Tillotson’s—dominated the colonial market and its supply of fiction should not serve as a framework for future investigations. [End Page 125]
Katherine Bode is Associate Professor of Literary and Textual Studies at the Australian National University. Her recent publications include Reading by Numbers (2012) and the co-edited collection Advancing Digital Humanities (2014). Her essay in this issue of VPR canvasses ideas developed and applied in her forthcoming book A World of Fiction.
1. Identifying and reprinting relevant or interesting content was a central part of the nineteenth-century newspaper editor’s job. Although “scissors-and-paste” journalism was discussed in a pejorative sense, there was no “clear professional consensus … about how much copying was too much, or how soon was too soon to reprint another paper’s material” (Nicholson, “You kick the bucket,” 275). For discussion of reprinting as a feature of nineteenth-century Australian journalism, see Kirkpatrick, Sworn to No Master.
2. Numerous studies have shown that the practice was increasingly formalised as the nineteenth century progressed, from unauthorised “borrowings” by individual editors to mutual systems of “exchange” and later to fiction syndication companies. Key scholarship for America includes Johanningsmeier, Fiction and the American Literary Marketplace, and McGill, American Literature; for Britain, Donaldson, Popular Fiction, and Law, Serialising Fiction; and for Australia, Hilliard, “Provincial Press,” and Morrison, “Serial Fiction.”
3. Prominent digital humanities projects have investigated reprinted content. Bob Nicholson used keyword searches to identify reprinted American jokes and slang in British newspapers (“Looming Large”); more ambitiously, the Viral Texts Project employs a text reuse discovery algorithm to identify reprinted passages in multiple genres (“Viral Texts: Mapping Networks of Reprinting in 19th-Century Newspapers and Magazines,” http://viraltexts.org). See also Cordell, “Reprinting,” and Smith, Cordell, and Mullen, “Computational Methods.”
4. Most of the titles in this study (98 percent) are extended due to serialization (publication over at least two, usually many more, issues of a newspaper); some titles (2 percent) appeared in a single issue, while comprising 10,000 words or more (often much more, with some titles identified in this project amounting to more than 60,000 words in a single, usually a Christmas, newspaper edition). This project also discovered over 7,000 short fiction titles, completed in a single issue and also extensively reprinted. Where these single-issue stories suggest incidental publishing and reading, I focus on extended fiction because it implies a more deliberate sourcing and reception of fiction. Although extended fiction appeared in Australian newspapers as early as 1828, I concentrate on the period from 1865 to 1899, when over 98 percent of the titles discovered in this project were published. The full dataset used in this study, including individual figures, is available at https://katherinebode.files.wordpress.com/vpr_data-and-figures.xlsx. I am in the process of creating a searchable and downloadable database for all of the fiction discovered, which I aim to release in late 2017.
6. Such effects include errors introduced in the OCR rendering of searchable text; quality and zoning issues implicated in the composition of digital collections; and the modeling of contents by search and relevance ranking algorithms and other features of collection interfaces. For early work in periodical studies highlighting the partiality of mass-digitized collections, see Mussell, Nineteenth-Century Press, and Solberg, “Googling the Archive.”
7. Library of Congress, National Digital Newspaper Program, http://www.loc.gov/ndnp/; findmypast, British Newspaper Archive, http://www.british-newspaperarchive.co.uk/; Europeana Newspapers, “Europeana Newspapers is Making Historic Newspaper Pages Searchable,” http://www.europeana-newspapers.eu/.
8. This estimate is achieved by comparing titles digitized by Trove with those listed in the three editions of the Australian Newspaper Directory (Gordon and Gotch, Australian Newspaper Directory, 1886, 1888, 1892) and averaging the results. Of the newspapers published at least once a week (the category predominantly digitized by Trove), Gordon and Gotch identify 749 in 1886, 868 in 1888, and 647 in 1892. In contrast, at the time I stopped harvesting fiction, Trove contained 142, 161, and 171 newspapers operating in those respective years, equivalent to 19 percent, 19 percent, and 26 percent of those in operation, or an overall average of 21 percent. (Calculations for 1892 exclude West Australian and Tasmanian newspapers, as these colonies were not included in the Australian Newspaper Directory I consulted for this period.) For discussion of gaps in analogue holdings of Australian newspapers, see Morrison, “Archaeology of Australian Colonial Newspapers” and “Retrieving Colonial Literary Culture.”
9. For instance, Johanningsmeier identifies 15,205 (2,226 daily and 12,979 weekly) papers operating in America in 1899 (Fiction and the American Literary Marketplace, 17). This can be compared with 647 daily and weekly newspapers available in Australia in 1892 (see footnote 8) or even 740 when fortnightly and monthly newspapers are included (Gordon and Gotch, Australian Newspaper Directory, 1892).
10. Based on a comparison with the three Gordon and Gotch indexes, metropolitan newspapers were almost twice as likely to be digitized as provincial papers for most colonies. The exceptions are Victoria (where an average of 20 percent of metropolitan newspapers are digitized, as opposed to 25 percent of provincial newspapers) and South Australia (where the average is 62 and 71 percent, respectively). These averages are the extreme ends of another effect of the digitization process: newspapers from colonies with smaller populations (Queensland, South Australia, Tasmania, and Western Australia) are more represented in Trove than those from larger colonies (New South Wales and Victoria). For the years covered by the Gordon and Gotch directories, an overall average of 40 percent of newspapers from the [End Page 127] former group are digitized, compared with 23 percent from the latter. As described in footnote 8, of the newspapers listed by Gordon and Gotch, a higher proportion (26 percent) are digitized for the final year (1892) than for the previous two years (with 19 percent coverage for 1886 and 1888). While this result indicates increased coverage over time, the opposite trend is suggested by comparing Trove’s holdings with Rod Kirkpatrick’s totals for provincial newspapers in the colony of New South Wales. In this case, 100 percent of titles identified by Kirkpatrick in 1850 are digitized, decreasing to 71 percent in 1860, 37 percent in 1870, 33 percent in 1880, and 28 percent in 1890 (Kirkpatrick, Country Conscience, 30). I am strongly inclined to believe the trend indicated by comparison with Kirkpatrick’s study for three reasons. First, overrepresentation of early titles makes logical sense: given the overall substantial growth in the number of newspapers operating in Australia across the nineteenth century, digitizing a relatively small number of titles from early decades captures a relatively large proportion of the historical total (the same logic applies to the smaller/larger colony comparison discussed above). Second, despite the much smaller range of newspapers listed, Kirkpatrick’s longitudinal span makes it a more reliable indicator of trends over time. Finally, and most convincingly, the increased proportion of newspapers digitized in 1892, based on the Gordon and Gotch comparison, is attributable to a particular combination of historical factors (a significant economic depression in Australia in 1890) and collection practices (a gradually increasing number of newspapers digitized by Trove). Whereas Trove’s strategy is designed to ensure a relatively consistent representation of periodicals in a context where the number of titles generally increases, the significant decline in Australian newspapers created by the 1890 recession, from 868 in 1888 to 647 in 1892, produces overrepresentation in this instance.
11. This method identified fiction by searching for terms used in their paratext and then harvesting and processing the results. The success of this approach is derived from the interaction of relevant search terms with four features of Trove’s interface: its page segmentation, relevance ranking algorithm, manual correction of title information, and Application Programming Interface (API). In contrast to Chronicling America and much of Europeana Newspapers, Trove segments or zones pages into articles, enabling the targeted searching of content. The relevance-ranking algorithm increased the likelihood of identifying fiction by returning articles to the top of the list of results in cases where the search term appears in the title (defined as the first four lines of text) and/or is recurrent in the article body. “Chapter” was the most successful search term employed in this project because it is frequently used both to introduce (appearing in the title) and to segment (appearing throughout the body of) fiction instalments. Given the focus on terms used in article titles, manual correction of this information—to 99 percent accuracy—ensured [End Page 128] that OCR errors did not significantly affect search results. Finally, Trove’s API enabled me to export search results wholesale and in a form amenable to automatic and semi-automatic data processing. For in-depth discussion of the challenges of automatically identifying and harvesting fiction from digitized historical newspapers, see Bode and Hetherington, “Retrieving a World of Fiction.”
12. On the exclusion of supplements, see Brake, “Lost and Found.” The extent to which newspaper supplements were excluded from the collection practices underpinning Trove only became apparent through this study of reprinting. As discussed in the final section of this article, most of the fiction in provincial newspapers, because it occurred in syndicated supplements, appeared in a common sequence in multiple newspapers (often on identical, ready-printed sheets). Exploring provincial newspaper fiction, I discovered a number of instances where particular provincial newspapers appeared to publish an irregular number of titles—or only one—in a sequence. Occasionally, further investigation confirmed an irregular or one-off publication: the newspaper simply happened to publish the same story—often sourced from a popular British or American periodical—around the same time as the syndicate, perhaps coincidentally or perhaps motivated by that story’s syndicated appearance. Much more frequently, a newspaper’s apparently singular or irregular publication of the same titles as a syndicate turned out to be an effect of missing supplements. Less commonly, other factors meant that only limited instances of recurrent reprinting were discoverable by automatic means, particularly the limited availability of issues for digitization or poor quality microfilm.
13. “Distant reading” is a term coined by Franco Moretti in “Conjectures on World Literature,” now widely employed in periodical studies (see, for example, Liddle, “Reflections”). For a critique of “distant reading,” including its association with the decontextualizing strategies of “close reading,” see Bode, “Equivalence.”
20. Cordell and his colleagues note that their “analysis necessarily misses far more reprinted pieces than it identifies” due to gaps in Chronicling America and problems with their algorithm (Smith, Cordell, and Mullen, “Computational Methods,” E2). However, with respect to the relationship between Chronicling America’s holdings and the historical context they are investigating, “significant gaps” are acknowledged without details of their nature and extent (Smith, Cordell, and Mullen, “Computational Methods,” E1). [End Page 129]
21. When he departs from particular examples to characterise relationships within early American print culture more broadly, Cordell uses statistical descriptions of network models: for instance, “Brownlow’s Knoxville Whig has the highest betweenness centrality in this network” (Cordell, “Reprinting,” 432). More recent work steps away from this approach in seeking alternative ways of rendering network models so as to “discern the links that truly seem indicative of historical connections rather than data artifacts” (Cordell, “Two”). These alternatives include according additional weight to instances of reprinting with temporal or geographical proximity. Such attention to historical constraints is promising in terms of the capacity to use network models for exploratory purposes. But it does not overcome the broader problem I am articulating—one that Cordell also foregrounds—of constructing networks based on the highly partial datasets derived from mining mass-digitized collections.
23. In this method, millions of networks, containing all possible combinations of causes, are created to explore and contrast the range of possible dynamics.
24. For instance, I could use statistical measures of probability to extrapolate from observed instances of republication in the various types of newspapers to calculate the probability that the 50 percent of titles appearing only once in my sample would be republished if the approximately 80 percent of colonial newspapers not digitized by Trove were included. Alternatively, a “forest” network could be devised to explore the system dynamics that result when all possible causes of reprinting in nineteenth-century newspapers are considered.
26. Funding for digitization is declining even as large parts of the historical record remain un-digitized. Although Trove leads the world in its approach to digitizing historical newspapers, at the last federal election in 2016 the government announced large cuts to the National Library of Australia that are certain to impact the pace and perhaps even the possibility of future digitization efforts (see Hitch, “Trove”).
32. Ibid., 76.
35. Johnson-Woods, Index to Serials, 6. The newspapers referenced are the Australasian, Leader, Australian Town and Country Journal, Sydney Mail, and Queenslander.
37. Measuring the number of unique and reprinted titles within a calendar year is designed to limit the inclusion of instances where two newspapers published the same story from different sources. (This did sometimes happen but was much less likely to occur in the same year.) However, because this approach excludes a limited number of instances of reprinting that occurred in consecutive years (for example, when one newspaper began serializing a story in December of one year and another began in January of the next), it understates the incidence of reprinting present in the dataset.
39. I have discovered 70 titles shared by the Brisbane Courier and Queenslander, 102 by the Evening Journal and Adelaide Observer, 77 by the Telegraph and Week, and 31 by the Evening News and Australian Town and Country Journal. In contrast, I have found none shared by the Argus and Australasian; only one shared by the Age and Leader and by the West Australian Times and Western Mail; four shared by the Sydney Morning Herald and Sydney Mail (all prior to 1865); and six shared by the Adelaide Advertiser and South Australian Chronicle.
40. A list of over 100 titles, syndicated by Tillotson’s and discovered in this project’s analysis of Trove, is available as appendix 1 on my website at https://katherinebode.files.wordpress.com/2014/07/vpr_appendix-11.pdf.
44. See appendix 1 for details: https://katherinebode.files.wordpress.com/2014/07/vpr_appendix-11.pdf.
45. Johanningsmeier establishes McClure’s connection with Australia based on author correspondence and the publication of Mark Twain’s “The American Claimant” (secured at great expense by McClure in 1892) in the Age in that same year (Fiction and the American Literary Marketplace, 76). This expanded sample shows that Twain’s story was also featured in the Adelaide Observer and the Evening Journal in 1892, suggesting that McClure moved beyond the leading metropolitan papers in his engagement with the colonial market.
46. Titles syndicated by Watt in Britain and published the same month in Australian newspapers include James Payn’s “The Heir of the Ages,” Walter Besant’s “The World Went Very Well Then,” William Black’s “Wolfenburg,” and S. R. Crockett’s “The Grey Man.” Titles that appeared in Australian newspapers a month after their British appearance include [End Page 131] Black’s “Sunrise,” Besant’s “All Sorts and Conditions of Men,” and Robert Buchanan’s “Master of the Mine.”
47. For a list of authors, see appendix 2 on my website at https://katherinebode.files.wordpress.com/2014/07/vpr_appendix-21.pdf.
48. The top twenty most published authors in colonial metropolitan newspapers between 1865 and 1899 (with the number of publications in parentheses) are M. E. Braddon (64), Dora Russell (44), James Payn (33), Adeline Sergeant (31), B. L. Farjeon (29), Wilkie Collins (28), William Black (27), Ada Cambridge (25), George Manville Fenn (25), Margaret Oliphant (24), W. Clark Russell (24), Walter Besant (23), J. Monk Foster (22), Bret Harte (22), W. E. Norris (22), Henrietta Eliza Vaughan Stannard (writing as “John Strange Winter”) (20), G. A. Henty (18), Henry Herman (18), David Christie Murray (17), and F. W. Robinson (17). Cambridge, an Australian writer, and Herman, a British author, are the two exceptions on this list in that they are frequently serialised authors who were not associated with well-known syndicators in the sources I consulted. The latter instance is probably an omission of the sources rather than an actual lack of association, given Herman’s longstanding collaboration with Murray, who was syndicated by Tillotson’s and represented by A. P. Watt. Other writers associated with these well-known agencies and agents who were among the top forty most published authors in colonial newspapers during this period are Hawley Smart (16), S. Baring-Gould (15), Joseph Hatton (15), Margaret Hungerford (15), H. Rider Haggard (14), William Le Queux (14), Eliza Lynn Linton (14), Robert Buchanan (13), John K. Leys (13), Hall Caine (12), and Thomas Hardy (12).
49. Given the extent of anonymous publication in nineteenth-century Australian newspapers, it is likely that the actual count of British authors was greater (see Bode, “Thousands of Titles”). Figure 4 only includes those whose identities have been verified.
53. Of the eight newspapers centrally involved in reprinting fiction within the colonies, five are the most prolific metropolitan publishers of fiction in this study: namely, the Queenslander (323 titles), Leader (299), Adelaide Observer (273), Evening News (255), and South Australian Chronicle (251). The other newspapers I have identified as heavily involved in reprinting—the Evening Journal, Telegraph, Week, and Brisbane Courier—are in 9th, 10th, 11th, and 13th place, respectively.
55. As in figure 1, rates of reprinting among provincial newspapers are based on the number and proportion of non-unique titles per year. While this [End Page 132] approach is useful for the comparison, it understates the extent of reprinting among provincial newspapers, which tended, more than for metropolitan cases, to occur in consecutive years.
56. I have identified 124 titles published by both newspapers. These are predominantly short serials (across two or three issues) that appeared most frequently from 1878 to 1882.
57. For the Hay Standard, just over 500 kilometres from Goulburn, publication typically began two weeks after it occurred in the Goulburn Herald; for the Cootamundra Herald, just under 200 kilometres from Goulburn, publication occurred in the following week. Altogether, the Goulburn Herald published twenty-seven titles with the Hay Standard and sixteen titles with the Cootamundra Herald.
58. For instance, “Wynnum White’s Wickedness” appeared in at least nine provincial newspapers in 1895 (Armidale Chronicle, Bathurst Free Press, Gympie Times, Morwell Advertiser, Nepean Times, Port Macquarie News, Richmond River Herald, Traralgon Record, and Western Herald). Hennessey also syndicated “An Australian Bush Track” in 1896 (Bathurst Free Press, Gympie Times, Telegraph, Week, and Western Grazier); “The DisHonourable: A Mystery of the Brisbane Floods” in 1895 and 1896 (Barrier Miner, Bathurst Free Press, Morwell Advertiser, Richmond River Herald, and Traralgon Record); “The Mystery of Sea-Cliff Towers” in 1897, 1898, and 1899 (Bendigo Independent, Goulburn Herald, Murrurundi Times, and North Queensland Register); and “The Bells of Sydney” in 1899 and 1900 (Clarence and Richmond Examiner and Ulladulla and Milton Times). Sampson Low, London, published the first three stories as books in 1896.
60. Price Warung’s “An Endorsement in Red” appeared alongside Hennessey’s “The Mystery of the Sea-Cliff Towers” in Hennessey and Harper’s 1898 Christmas annual and was subsequently published in the Western Grazier in 1898.
61. Lucy Sussex describes an abortive attempt by Borlase to establish a syndication agency for original local fiction that was announced but never appeared. Sussex, “Bobbing around.”
62. Further details, including specific titles, authors, and dates of publication, are published on my website as appendix 3, https://katherinebode.files.wordpress.com/2014/07/vpr_appendix-3.pdf.
63. Book sellers and lending libraries were scarce in the Australian colonies (Johanson, Study of Colonial Editions, 213; Eggert, “Robbery under Arms,” 134) and local literary periodicals were short-lived (Stuart, Nineteenth-Century Australian Periodicals, 1). Though imported literary magazines and journals were present and popular, they were significantly more expensive and less prevalent than the “large, vigorous and thriving” local newspaper press (Morrison, “Serial Fiction,” 308). [End Page 133]
65. Cameron, Laing and Co. published twenty-six Australian titles from 1880 to 1884, plus an additional two that were almost certainly Australian (“Kitty Dunolly, My Schoolmate: A Victorian Sketch,” by G. E. C., and “Bonshaw: A Moreton Bay King,” by Magnus Badge). In contrast, George Robertson, the most prolific local book publisher, published nine Australian novels between 1860 and 1889 and twenty-two in the 1890s (Bode, Reading, 44, 49).
66. National Library of New Zealand, “Papers Past,” http://paperspast.natlib.govt.nz/cgi-bin/paperspast. Titles published by Cameron, Laing and Co. that appeared in New Zealand newspapers such as the Tuapeka Times, Hawera and Normanby Star, and Waikato Times include “Denis Devine,” “In the Folds of the Serpent,” “The Mystery of Major Molineux,” “Bonshaw: A Moreton Bay King,” “Marc Grecli,” “Dora Dunbar,” and “Days of Crime and Years of Suffering.” The involvement of an Australian syndication agency, S. & D. Reid, with New Zealand newspapers in the 1890s has been described by Harvey in “Sources of ‘Literary’ Copy,” but as far as I know, this earlier cross-Tasman connection has not been noted previously.
68. I have decided, on the balance of evidence, that syndicate 1 ceased operations in 1892, but it is also possible that it continued, publishing fiction I have allocated to syndicate 6. In support of the first interpretation are the different newspapers involved (more than half of the periodicals associated with Cameron, Laing and Co., up to and including 1892, were no longer publishing the same fiction in 1893), the different locations of these newspapers (syndicate 6 worked mostly with Victorian rather than New South Wales publications), and the different type of fiction published (syndicate 6 published a large proportion of titles by unknown authors). Supporting the second interpretation is the involvement of similar authors. A number of the Australian writers published by syndicate 6 were previously associated with Cameron, Laing and Co., including James J. Wright (writing as Ivan Dexter and Captain Lacie), Kenneth Hamilton, Harold M. MacKenzie, and Atha Westbury. A number of the same newspapers were also involved. Almost half of the periodicals previously associated with syndicate 1 appear in syndicate 6, although two thirds of the newspapers in the latter were not previously aligned with Cameron, Laing and Co. A change in ownership might explain such dramatic shifts in publishing and business practices. But Cameron, Laing and Co. was acquired by S. & D. Reid in 1888, so the timing seems to discount this explanation (Harvey, “Sources of ‘Literary’ Copy”).
70. For example, Fenn’s “Commodore Junk,” Russell’s “The Frozen Pirate,” and P. T. Barnum’s children’s serial “My Plucky Boy Tom” were published concurrently by this syndicate. [End Page 134]
71. The difference in timing relates to the day of the week on which these (typically bi- or tri-weekly) provincial newspapers were issued.
72. Whereas syndicates 7 and 10 appear only to have traded in partly printed pages, syndicates 6, 8, 9, and 11 offered a combination of partly printed pages and flexible reprinting.
73. In Fiction in the American Literary Marketplace, Johanningsmeier provides a long list: “The American Short Story Company, the American Press Company, Frank Carpenter’s Newspaper Syndicate, the International Syndicate of Baltimore, the Albert Bigelow Paine Syndicate, Syndicate Exchange, the Lorraine Literary Press Association, the Authors’ Co-Operative Company, and the Wilson Press Syndicate” (96). Indeed, he notes the difficulty of investigating even the major American syndicates—Bacheller’s and McClure’s, which were “ubiquitous in the Anglo-American literary publishing world of the 1880s and 1890s”—due to the “paucity of available manuscript and secondary materials” (67, 71).
74. For syndicate 7, the inclusion of four serial stories by Bertha Clay (an author name of disputed origin but strongly associated with fiction syndication in America) could indicate an American company or an Australian agency that actively acquired fiction from American sources. Alternatively, its inclusion of advertisements for colonial companies in its partly printed pages (for instance, for “Australian Explosives” and a Melbourne dentist) could indicate a locally based agency or an overseas syndicate producing partly printed pages specifically for the colonial market and seeking advertising revenue in that context. For syndicate 11, the mixture of international fiction and the inclusion of miscellaneous American materials could suggest an overseas company providing general material for the international market or a local company extracting such material from international newspapers for colonial publication.
75. Challenging the view that local publishing did not occur in Australia until the late twentieth century was also a key focus of my 2012 book, Reading by Numbers (27–103).