Editing a Paper

Counting Newspapers

Before we can begin a computational study of what nineteenth-century editors put in their newspapers, we must account for the the editorial and curatorial decisions that shaped the collections of digitized newspapers we use today. We observe the activities of writers, editors, and compositors through a long chain of choices made long after a newspaper was set in type. In this section we build on data archeologies of digitized newspaper corpora by authors such as Paul Fyfe (2016), Ryan Cordell (2017), and Benjamin Lee(2021), as well as the Atlas of Digitised Newspapers and Metadata written by M. H. Beals et al., to “advocate for the importance of autoethnographic approaches to documenting a cultural heritage dataset’s construction from a humanistic perspective” (Lee 2021). Rather than repeat the investigative work into collections like Chronicling America from this existing literature, we triangulate between multiple historiographies and datasets to complement and enrich those archeologies.

We begin with a set of seemingly simple baseline questions about how we might even count the newspapers we study, which requires us to trace the cocentric circles the Stanford Literary Lab names “the published, the archive, and the corpus” (Algee-Hewitt et al. 2016, 2). How many copies of newspapers did printers produce? How many survive to be collected by libraries and archives? Which newspapers are microfilmed? Which microfilms are digitally photographed and transcribed with OCR? Because of these later steps in the data-generating process, we try to be careful about what we can and cannot conclude from our observations.

To start this inference process, we describe how we interpret the data in the Library of Congress’s U.S. Newspaper Directory (USND). Since it covers only the current territory of the United States, the USND does not help us answer similar questions for British, Australian, or other newspapers we consider elsewhere in this book, but the analyses in this chapter provide models for analyzing newspaper metadata that could be applied to those other collections in the future.

Overall Counts

After cleaning data from the Library of Congress’s U.S. Newspaper Directory to estimate the number of distinct newspapers published in each year, we can see tremendous growth, from a few hundred papers in 1800 to over 12,000 by the end of the century, with only a slight reversal during the Civil War.

This data hews closely to Nord’s estimates, quoted above, for 1810 (“fewer than 400 newspapers”) and 1825 (“more than 800”), although they are significantly higher than the 900 he estimated for the year of Tocqueville’s visit. As scholars such as Louisa Trott have shown, the U.S. Newspaper Program’s catalog undercounts the newspapers that existed, particularly those published by minority communities, so we would expect so even our numbers likely undercount the historical reality, though it offers a more precise accounting than was previously available (Trott 2022). Our data also provides more precise accounts of newspaper proliferation in the latter half of the nineteenth-century, when the rapid growth of the medium becomes harder to track by hand.

If we looking instead at the original catalog data, our estimates would be over 10% higher still.

While these graphs do not rewrite existing scholarly accounts of newspaper proliferation, they do offer quantitative and precise evidence of that spread that particularly expand our picture of the mid and latter half of the century.

Digitized Newspapers

Much of what we want to know, however, requires us to look beyond the catalog to the contents of published papers. For the purposes of this book, we examine digital photographs of newspaper pages and read—both computationally and closely—the digitized text. Elsewhere in this book, we consider newspapers from beyond the United States; for most of this chapter, we focus on four digital collections with digitized papers that overlap with the USND.

Given what we know about the concentric circles of the published, the archive, and the corpus, we might wonder how likely it is any given paper will be digitized? This probability varies by year, place of publication, language, and other features. To assess it, we compare the proportion of papers in each year from the catalog with the proportion of various digitized collections that come from each year.

To make the different curves comparable, each one is normalized over the time interval, so that the area under each curve sums to 1. We see that both the Chronicling America and the Gale US collections oversample newspapers, relative to the catalog, from before 1870 and undersamples them after 1870. This is also true of the whole set of digitized newspapers, which includes data from ProQuest’s American Periodicals Series and the Internet Archive collections. Chronicling America, at least, still accurately reflects the fact that there are more papers published in years after 1870 than before; for Gale US, there is an absolute decline in the number of digitized papers after the Civil War, even though more papers existed after that date historically.

When drawing inferences from the digitized sample, then, we would not want to overgeneralize from the years before 1870. One solution might be to reweight the digitized sample to follow the distribution of years in the catalog. Unfortunately, the decision to digitize depends not only on the year but also on other factors, such as state, language, and frequency of publication.

The National Digital Newspaper program provided funding and infrastructure support from the National Endowment for the Humanities and the Library of Congress to state-level digitization efforts.1 States have applied to the NDNP at different times, and some states have received more rounds of funding than others, such that their newspapers are better represented in digital collections. For example, we can see that the proportion of newspapers in the catalog from each state or territory does not match the proportion of newspapers in the digitized sample.

Displaying the same data in another way, we can see that the proportion of papers digitized varies greatly among states, and is often quite far from the overall rate of digitization of 5% indicated by the vertical line.

Other features of newspapers are distributed unevenly in space and time. As we will see below, the frequency with which a newspaper is published leads to different patterns of reprinting. Overall, the digitized collections overrepresent daily papers, and different states had different numbers of daily papers compared to other publication frequencies.

A newspaper’s language has an even bigger effect on what it prints and reprints. Although the overall digitized sample is fairly representative of the population, with 4.5% non-English papers, the proportion of non-English papers both in the catalog and among digitized papers varies by state. In Minnesota, the digitized sample overrepresents the proportion of non-English papers; in Hawaii, the digitized sample contains only 25% non-English, while the USND catalog contains 50% non-English.

Digitized data on the Black press provide an even starker example of the limits of a “representative sample”. Due to legal, economic, and social barriers to the emergence of the Black press, less than 1% of nineteenth-century newspapers, according to the USND’s coding, are African American, which aligns with Benjamin Fagan’s description of the “whiteness of these public archives” which can skew computational analyses and keyword search research alike. As Fagan argues,

we must be careful to acknowledge at the outset that any conclusions drawn from such archives apply specifically, and exclusively, to white periodicals, which cannot be made to stand in for all of early American periodical production. Otherwise, we risk conflating Americanness with whiteness. This is a slippage that the racial politics of digitization encourages us to repeat, which makes it all the more important that we reveal it in our teaching and scholarship (Fagan 2016, 12).

If we sought to understand this subset of nineteenth-century newspapers by digitizing even a small number of them, the sample becomes unbalanced in another way: digitizing only a single paper each from Oklahoma or West Virginia, or two papers each from Minnesota and Wisconsin, overrepresents the Black press in those states, at least as a proportion of digitized newspapers. Our approach, as we discuss below, is to explicitly model the process of selecting papers for digitization, rather than relying on the representativeness of any particular sample we might construct. This approach need not be one of resignation toward inequities, however. Following Rediet Abebe et al.’s argument that too much technical work “treats problematic features of the status quo as fixed,” we instead suggest that modeling how newspapers are selected for digitization can “serve as a diagnostic, helping us to understand and measure social problems with precision and clarity” (Abebe et al. 2020). Our goal is not to portray the current state of newspaper digitization as fixed and immoveable, but instead to offer a detailed account of what current collections include and do not, in hopes these conclusions might influence future digitization efforts.

Many features of newspapers vary across both space and time. In general, the proportion of daily papers increased during the nineteenth century, but at different times in different states. The small sample of digitized papers in many states leads to larger swings in the proportion of dailies among digitized papers.

How then should we generalize to the whole population from a non-representative digitized sample of ~5%? One approach from survey research is poststratification. Say that we conduct an opinion poll of residents of the United States. In addition to asking the respondents about the topic of interest—what candidate they indent to vote for or how many pets they have—we also ask for some demographic data. If we receive fewer responses from women—or from residents of Pennsylvania—than we would expect based on the proportion of women or Pennsylvanians in the population, we could estimate poll responses for those subpopulations and then combine them according to those known population proportions. Using (for better or worse) 2020 U.S. Census categories of males and females, we could collect poll responses for men and multiply them by 49% and poll responses for women and multiply them by 51%.

We could also stratify men and women into further subpopulations by age and state of residence. (Stratification by age might also lead us to limit the population of interest: most polls don’t attempt to estimate the opinions of children; election polls try to concentrate on the opinions of voters.) If we add education, party affiliation, race, and religion, the number of different bins is multiplied considerably. Many bins may contain no respondants to our poll. If we get a thousand responses to a nationwide poll, we might have no responses from male Democrats from Wyoming who are 40–65 years old. Responses from other men, Democrats, Wyoming residents, and 40–65-year-olds, however, might help us estimate how those categories would interact. More specifically, we can go beyond counting responses in a fixed number of bins to estimating a statistical model of the poll response given observed categories like state, age, etc. One widely-used version of this technique is multilevel regression with poststratification (Mr. P, i.e., “Mister P”, Gelman and Little, 1997).

Consider, for the sake of concreteness, the variables in the USND indicating the year (100 values) and state (52 values) of publication, as well as Boolean (two-valued) variables for “African American”, “daily”, and “non-English”. If we could set each of these variables independently, we could potentially sort newspapers into one of 41,600 possible bins.

In reality, these variables are correlated. To take a simple example, different states acquired printing presses, and thus newspapers, in different years. The USND catalog, therefore, shows that only about 10,000 of these bins, or 25%, are occupied. Of those bins in which the catalog records newspapers, we only have digitized data for about 6000 or 59%.

On the other hand, 59% is greater than 5%. That is to say, if the phenomena we’re interested in—such as the density of reprints or advertisements, or number of pages or images or recipes published in a newspaper—can be explained by the five variables listed above, then we have observations for more than half of the actually possible settings of those variables, which is greater than the overall 5% sample at the newspaper level. Of course, we might decide that other variables such as city (not just state) or political affiliation might be useful, thereby increasing the number of bins.

The histogram shows not only a large number of bins with no digitized newspapers but also many bins with only one or two observations. In the rest of this chapter, we will describe methods for using evidence from more densely populated bins to help us draw inferences about sparsely populated or empty bins.

Newspapers and Political Partisanship

First, during the first half of the nineteenth century the tie between newspapers and political partisanship in the US solidified, codifying editors’ roles as local political advocates and agitators. In the United States, these political roles were filled locally. Most towns large enough for one newspaper in fact had (at least) two, one for each of the major political parties, and possibly others advocating for particular political goals, such as abolition. As Pettegree notes, from the 1830s US papers “mixed trenchant political commentary with reports of lurid crimes and local scandal” with “little pretense of objectivity” (Pettegree 2021, 634). For a vivid example of these political affiliations, we might look to Brownlow’s Knoxville Whig and Independent Journal, which was named—at various times and locations during its long history under one editor, Parson William Gannaway Brownlow—the Tennessee Whig, the Whig, the Jonesborough Whig, the Jonesborough Whig, and Independent Journal, the Knoxville Whig and Independent Journal, and Brownlow’s Knoxville Whig, and Rebel Ventilator. As noted in the National Digital Newspaper Project’s biography of Browlow’s papers, “Brownlow’s speeches and publications drew both attention and anger” throughout his career, such that he was sued and shot multiple times during his life; was arrested and charged with high treason against the Confederacy; was provided by “the federal government…with a press, some type, $1,500, and a government printing contract” when he returned to Knoxville in 1863; and “was a force in the convention that abolished slavery in Tennessee and that led to the creation of a new state government,” to which he was elected as governor immediately after the war (University of Tennessee and National Endowment for the Humanities n.d.).

The motto on the masthead of Brownlow’s Knoxville Whig sums up this political fervor neatly, declaring the paper “Independent in All Things—Neutral in Nothing.” This was a relatively common motto for newspapers in the period, and its oft-repeated notion of “independent” is complicated by the fact that many papers who used this motto, Brownlow’s included, were directly propped up by political parties and interests.3 Editors’ political allegiances could be complex and shifting, and the volatile business of newspaper production saw papers founded, disbanded, sold, and merged in ways that affected their political voice. The Elk County Advocate (Ridgway, PA), for example, was founded as an independent paper in the 1850s but was by the mid-1860s clearly Democratic in its endorsements and advocacy. When the paper was sold to a Republican editor in 1868, the title stayed the same but the numbering started over, so the November 20, 1868 issue is identified as Volume 1, Number 1. The nearby Clearfield Republican (Clearfield, PA) “had nineteen owners, five titles, and…changed its politics four times” through the nineteenth and early twentieth century,(Swoope 1911) and “Despite its name, the Republican for most of its life was an organ of the county’s Democratic (majority) party and was a robust example of Copperhead politics during the Civil War” (Libraries and National Endowment for the Humanities n.d.). Perhaps because of these complexities, the partisanship of the nineteenth-century press contributed substantially to its expansion, as political operatives, editors, and readers alike sought to ensure their views were represented to the public.

That the newspaper, as a medium, was partisan does not mean it was only, or even primarily, political. Among other insights, Viral Texts’ analyses demonstrate the radical generic diversity of most newspapers, even those fervently devoted to political advocacy. As Andie Tucher notes, “Many papers came (or were brought) to town expressly to counter the opposition parties or factions or splinters that were already there, but in the intervals between elections, even the most partisan papers often had a haphazard, milk-mild, ‘something-for-everyone’ air, closer in spirit to a magazine than a newspaper” (Tucher 2010, 396). The reality, we would suggest, is that separating newspapers from magazines on the basis of tone, content, or even format alone can be nearly impossible in this period, as the two genres overlapped and bled into each other.

As editors complained through the widespread reprinting of the listicle “Editing a Paper” (~158 reprints) trying to meet so many readerly needs through a single medium often proved impossible:

Many newspapers, in other words, aimed to be simultaneously partisan and playful, educational and entertaining. So while newspapers typically advocated explicitly for particular political parties and causes, in ways that can shock modern readers expecting journalistic impartiality, this did not prevent them also publishing recipes, fiction, poetry, and any manner of other texts of more casual interest to their readers.

The Technologies of Newspaper Production

Regardless of readerly and political interests, the dramatic expansion of the newspaper during the nineteenth century could not have happened without concurrent technological changes that made it possible to produce such a large quantity of print at prices affordable to many readers. First among these was the development of reliable methods for making paper from wood pulp rather than rags. For most of print history, procuring sufficient paper constituted the highest expenditure for the printer on any given job, and thus largely determined the price for consumers. Rags of sufficient quality to make good paper were ever in short supply, and the search for a cheaper alternative was continuous throughout the print era. This need can be illustrated by the Mountaineer, from the new and relatively-isolated Salt Lake City. On 20 July 1861, the paper wrote,

We are pleased to learn from our friend, A. C. Pyper, that there is a good prospect for a fresh supply of paper. Much means and labor have been expended in endeavoring to manufacture and introduce into the market this much needed article. We are sorry to say that we will be compelled to sustend (sic) operations in our office for a week or two for want of the material. It is to be hoped, however, that before long we will renew our publication without interruption.

Despite their hopes, however, the Mountaineer never resumed operations. We cannot know for certain, of course, that lack of paper caused its ultimate demise, but this blurb demonstrates how central this resource was to newspaper production, and how valuable it was, as late as 1861, in areas without established paper manufacturing facilities. Similarly, the Opelousas Courier in Louisiana printed multilingual issues (in English and French) on wallpaper at times during 1863 and 1864, when broken supply chains during the Civil War made more conventional papers impossible to obtain.

Printers’ need for more and cheaper paper, in part to print more and larger newspapers, drove paper manufacturers towards industrialization. From the early decades of the nineteenth century, paper mills developed ever bigger and faster steam-driven machinery for creating paper from rags, and in larger rolls or sheets than had been possible when paper was made using hand-held forms. Through the same decades, nineteenth-century paper makers experimented with methods for converting plant fiber, such as wood pulp, into paper. These methods were not perfected until the 1860s, when wood-pulp paper helped further drive down the cost of newspaper production (Weeks 1969, 234–36; Pettegree 2021, 634). The introduction of cheaper resources, then, combined with new industrial methods for processing it, made a material substrate for the newspaper capable of supporting the medium’s rapid expansion throughout the nineteenth century.4

Industrialization transformed paper-making and then printing itself, as iron and then steam-powered rotary presses made it possible to print sheets at speeds unheard of during the hand-press period (which spanned from Gutenberg’s press, developed around 1439, until the mid 1840s). Inventors had experimented with rotary presses since the late eighteenth century, working to develop a press in which paper feeds into a cylindrical printing mechanism that rolls continuously, rather than a platen that must be pressed and lifted for each new impression. The first successful rotary press was patented in by Richard March Hoe in New York in 1843, and quickly spread around the world. As described in R. Hoe & Company’s 1873 catalogue, the precise speed of printing depended on the number of cylinders in a given machine, from two up to ten. Their “Rotary Perfecting Newspaper Press” was “designed to work exclusively from stereotype plates” and could “print…both sides of the sheet at one operation.” This machine was designed specifically for working newspaper offices and could be ordered in a “Two-Feeder” model, which required two operators and could “print, per hour, from 4000 to 5000 perfected eight-page sheets of ordinary size, or will print and cut from 8000 to 10,000 perfected four-page sheets or ordinary size.” The “Four-Feeder Machine” could double that output, but required more operators (R. Hoe & Company 1873, 13). We should not imagine, however, that every small-town newspaper was printed on a rotary press, at least not immediately. The majority of US papers did not require the speed or scale of such machines, nor could they support the creation of the cylindrical stereotypes required to print on them. The introduction of rotary steam presses in large urban markets reshaped the overall press economy, making hand presses more readily available and affordable to printers in smaller markets. Taken together, these advances in paper production and printing aided the newspaper’s rapid proliferation during the middle decades of the nineteenth-century.

Even with these technological advances, the newspaper was a massive commitment of time and labor in small print shops. Many small-town and rural newspapers were produced by a very small team comprising an editor (who might also assume some production duties) and a few compositors or press workers, who likely produced other jobs alongside, or even ahead of, the newspaper. As a popular selection from the 1870s claimed, “a newspaper the size of the [Poughkeepsie] Eagle,” a typical 4-page paper, used 600,000 pieces of type, “the actual number of bits of metal arranged and rearranged every day in preparing a newspaper. Of course, a good deal of material stayed in type from day to day, including advertisements, mastheads, subscription notices, and editorial statements, so the 600,000 figure might be slightly exaggerated. The core sentiment of this selection, however, resonates: the paper was a huge and unending undertaking simply to compose. As with rotary presses, the invention of Linotype and other hot metal typesetting technologies in the late nineteenth century would, eventually, transform newspaper composition, but neither universally nor immediately.

As this book will show, a similar mix of older and newer technologies facilitated the sharing of texts across the newspaper network. We might imagine, for instance, that the invention of the electric telegraph would completely overhaul information sharing during the nineteenth-century. The Associated Press, for example, was founded as a consortium of New York newspapers in 1846 and used telegraph lines to become the dominant conduit for news sharing in the United States, “quicken[ing] the pace of the news cycle while intervening in the already intricate geographies of newsprint” (Gitelman and Mullaney 2021, 177). As cities connect through the expansion of telegraph lines, we do find in the Viral Texts data that the exchange of some kinds of stories—i.e. “hard news”—shifts, among some partners—i.e. mostly urban editors—to the telegraph. However, due to the expense of telegraph transmission, the vast majority of newspaper texts continued to circulate primarily through the mail until century’s end, and even those texts that first circulated by telegraph were then picked up and transmitted through other means to newspapers who lacked access to telegraphic news. Through the end of the nineteenth-century, in fact, the transmission of most newspaper texts—besides reports of things such as battles or elections—seems to remain primarily postal, rather than telegraphic, even as telegraph lines become a literal network across the US landscape. In other words, no single technology reinvented the newspaper, nor were any technological changes instant or equally distributed.

Newspaper Exchange Networks

We cannot account for the proliferation of newspapers and newspaper content through the nineteenth century based only on political need or the capacity of printing technology. Neither of these helps us understand precisely which texts appeared in the newspaper or, entirely, how they found their way into print. Small newspapers—often staffed by just an editor and perhaps a few press workers—assembled dense, five- or six-column newspapers on a weekly or even daily basis. Such small teams could publish regularly because any given newspaper was only one node within a larger, networked “exchange system” through which newspaper material was shared in common, published and circulated for reprinting, commentary, and other forms of reuse elsewhere. Most individual papers, in fact, produced very little original content, but instead acted as aggregators, collecting interesting selections from their regional, political, or national exchanges and republishing them for local readers. Like modern blogs or news aggregators, such as Buzzfeed or Huffington Post, which use aggregation to produce regular content despite relatively small editorial staffs, nineteenth-century newspaper editors exploited systemic reprinting, which they called “selection,” to fill columns.

In exchanges, editors subscribed to each other’s publications and borrowed content promiscuously from each other’s publications. Ellen Gruber Garvey describes this labor: “On large papers, a special ‘exchange editor’ went through other papers for material,” a process that helped local newspapers expand “by yoking together scattered producers who shared labor and resources by sending their products to one another for free use” (Garvey 2012, 30–31). While copyright law protected domestic books from reproduction, the content shared through the newspaper exchanges was not protected under intellectual property law. Instead, periodical texts were considered common property for reprinting, with or without modification—much as articles, music videos, and other content are shared online today among blogs and social media sites. Indeed, US postal laws explicitly supported these practices. From the late eighteenth century until the 1870s, editors paid no postage to mail papers to each other. The postal system was intended as an information network, and distributing newspapers cheaply helped advance an idea of democratic access to knowledge.5 When texts were reprinted, they were sometimes marked as such, perhaps with a citation to the source paper, but just as often selections were republished without any attribution at all. When looking at a nineteenth-century newspaper page, certainly a lack of citation on any particular article or squib cannot be taken as evidence that piece was original to that paper.

The data we have collected through the Viral Texts project can help clarify just how much of a typical nineteenth century newspaper page was composed of reprints. Our database of reprint clusters not only indicates the texts that circulated widely, but can be used to estimate how much text, on average, was reprinted on the typical newspaper page. The graph below is based on the data from 1,949 distinct Chronicling America newspaper titles, and issues published between 1840–1899.6 The text on each page was compared against identified reprints from our project to see what percentage of the total text in each issue seems to have been printed in other newspapers prior to appearing on that page. In other words, if a segment of text belongs to a reprinted cluster, and that witness is not the first observed witness, it is considered reprinted material. In addition to tallying external reprints, this same method compares segments of text on each page against previous issues of the same title so that we can identity how much material, on average, newspapers recycled as boilerplate, advertisements, editorial statements, and other text that would have been kept in standing type for multiple issues.

Attitudes towards this ubiquitous scissors and paste journalism varied, with some commentators decrying editors’ lack of originality, or even theft, while others valued the range of information and entertainment selection practices brought to local or even isolated readers. Many editors insisted that their primary role was selection and that the primary value of their papers was the way they aggregated information from around the country. As the Fremont Journal (Ohio) claimed on December 29, 1854, “[d]eprive an editor of his exchanges, shut off his mails for a week, and you take away from him his very sustenance, and withdraw from him all that makes his paper interesting.” Newspapers, the Journal asserted, operate under a “general ‘reciprocity treaty’…so that each may assist the other in collecting intelligence and together circulate the vast amount of news, of politics and literature, that circulates through the thousand columns of the periodical press over the land.” Other commentators decried the lack of originality they saw evidenced by the exchange system.

The widely-reprinted listicle “Editing a Paper” laid out the dilemmas facing editors seeking to please readers, including those related to selection:

If we publish original matters, they damn us for not giving selections.If we give selections, people say we are lazy for not writing more and giving them what they have not read in some other paper.

The first reprinting of “Editing a Paper” identified by the Viral Texts project appears in the Big Blue Union of Marysville, Kansas, but even here an editorial preface begs ignorance about the list’s origins, claiming to only know that it has been “going the rounds of the papers. If we knew in what paper it first appeared,” the editor continues, “it would afford us pleasure to give the writer due credit” (11 July 1863). This piece and its preface illustrate much about editors’ and, presumably, readers’ attitudes toward reprinting and the texts which circulated through the newspaper. While they express some concern about authorship and proper credit, circulation and selection ultimately matter more. Being able to credit the author would “afford us pleasure,” but not knowing the author or origin will not prevent an editor from republishing the piece.

“Within the last few days we have added to our Exchange list the following Periodicals,” the Edgefield Advertiser(16 September 1841) wrote to introduce a series of paragraphs outlining the virtues of four new additions to its exchanges. It was not unusual for papers to publish a brief account of why another publication had been added—or in some cases removed—from its list, though this example is unusually thorough. The publication of this article signals that the Advertiser’s readers cared about their local paper’s sources and might even seek to personally subscribe. The Advertiser’s laudations can help clarify what editors valued in each other’s work, and in particular these paragraphs demonstrate that selection and editing were marked as skills equal to editorializing. Of the Southern Botanical Medical Journal(Forsyth, Georgia), the Advertiser writes “The editorial and selections, as far as we are able to judge from a hasty perusal, are good,” praising equally both the original and reprinted elements of the publication. Similarly, the Western Farmer (Detroit, Michigan) is praised for “its editorial and selections…from which our Agricultural friends might gain knowledge,” while the Southern Planter (Richmond, Virginia) is summed up as “well edited, and bids fair to be useful to Planters, Farmers, &c.” This article does not write about selection and editing as secondary or subordinate to original writing, but as an equal guarantor of a publication’s quality.

While the “Editing a Paper” sums up editing as a thankless task, other examples of meta-reflection signal that participants in the exchange system valued discerning eyes, deft scissors, and careful composition, and even pushed back against contemporary rhetoric that sought to promote hierarchies within the system. Consider this brief article about the New York Evening Mirror from the Cadiz Sentinel (19 February 1845):

There is no paper on our exchange list we read with more exquisite pleasure than the N. Y. Evening Mirror. Willis is one of the few original and intellectual writers of the age, and wields a bold and vigorous pen.

In many ways, this article’s laudatory opening seems to contradict what we have thus far written about originality and reprinting, praising the “original and intellectual” writer Nathaniel Parker Willis—the brother of Fanny Fern, who we will discuss in Chapter 5—who was hired in 1844 by the Mirror’s editor George Pope Morris to help reimagine and revitalize that paper. We might contrast this description of Willis’ “bold and vigorous pen” with the Houma Ceres editor, described earlier in this chapter, who apologized for being “not…very distinguished as a ‘knight of the gray goose quill.’” Here an urban paper and literary writer seem to be held up for special esteem, distinguished in a way a rural newspaper editor cannot be. However, the Sentinel immediately pivots from its praise of Willis to describe the Mirror as “the representative of the ‘upper ten thousand,’” a phrase coined by Willis himself to describe the 10,000 wealthiest residents of New York City, and which came to be used as a rhetorical placeholder for the American gentry.7 As the representative of the elite, the Sentinel allows that the Mirror “of course contains much reading that would be of little interest to the individual who penetrates the forest with his axe, his musket and his dog. But the class of readers for whom it is intended,” the Sentinel continues, “may certainly be proud of the brilliancy of their organ.” By identifying “the class or readers for whom” the Mirror is intended as only the “upper ten thousand,” the Sentinel implicitly claims Willis’ “bold and original pen” has a limited reach. If it cannot address the individual in the forest with his axe, musket, and dog, then it cannot, perhaps, address most of the nation.

From here, the Sentinel notes,

“The Mirror of the 10th inst. we perceive has republished an article from the Sentinel on social parties, and introduces it with the following remarks:8

NOTIONS ON SOCIETY.—The Cadiz Sentinel is enlivened with a vigorous argument as to the character of the evening parties in the society of that Ohio village. The following is the deprecatory plea for gayety of the sinner-combattant. [sic]

After praising the “original and intellectual” qualities of writing in the Mirror, the only original prose the Sentinel actually quotes is a single sentence, and one that introduced a reprinted text. What prompts the Sentinel to comment on the Mirror, in other words, are the operations of the exchange network that brought their writing from Ohio and to New York City readers. For all the literary ambitions of the Mirror and Nathaniel Parker Willis, this reprint evidences that print culture does not simply emanate from the metropolis to benighted rural villages, but that through the exchange system urban papers draw on rural writing as well. By explicitly describing the Mirror’s “class of readers” and then calling attention specifically to the way it introduced a reprint from the Sentinel, this article prompts its readers (and us) to consider the urban paper’s subtle condescension, which is belied by its very need for exchange material from papers like the Sentinel.

Approaching the newspaper computationally, focused on reprinting practices, Going the Rounds offers one answer to the “difficulties of tracing editorial practice” Jim Casey and Sarah H. Salter identify in a recent edition of American Periodicals. We concur that “Editors were not merely stand-ins for authors, printers, or publishers” and that understanding editorship requires “the analysis of historically situated editorial practices” (Casey and Salter 2020, 102). It is perhaps strange, then, that we approach editors’ work—and, we would argue, the entangled work of compositors and press workers—through the ahistorical phrase “virality.” As we argue in the next section, however, we employ that frame to destabilize common literary-historical tendencies—our own as well as other scholars’—to center authorship and originality, even when reckoning with a medium that deemphasized both qualities.

Notes

1. For more on the NDNP and funding, see Ryan Cordell, “‘Q i-Jtb the Raven’: Taking Dirty OCR Seriously,” Book History 20 (2017): 188–225.

   Return to note reference.

2. This is a simplification by the USND: newspapers did change their frequency over time, so while catalogs often assign one frequency category to particular titles, the historical realities of production were often less regular. Ostensibly weekly papers would sometimes issue more frequently, such as during election season when they were actively stumping for their party’s candidates, while ostensibly daily papers would sometimes issue more sporadically, as happened to many papers during the Civil War when supplies ran low or armies interfered with routine. As long as a paper did not change its title, however, the USND catalog does not reflect changes in frequency, place of publication, or other features. In the next section, we will directly model changing properties of newspapers, such as their length in pages or words.

   Return to note reference.

3. Brownlow’s Knoxville Whig continues after Brownlow himself leaves in 1866 to become governor, when his son takes over the paper. They sell the paper in 1869 to others who remove Brownlow from its name;, but Brownlow returned to the paper in the mid-1870s, when it is called the Knoxville Whig and Chronicle. In other words, even a paper like this, so strongly tied to one person, persists after that person leaves it, albeit under a new name, and the paper retains “Whig” in its title long after the Whig party faded in the U.S. The actual politics of the paper continue in their fervency, backing new parties and politicians, even as the official name of the paper is highly personal and idiosyncratic.

   Return to note reference.

4. For the most comprehensive account of paper production and use in nineteenth-century America, see the work of Jonathan Senchyne, in particular (2017) and (2019). We thank Professor Senchyne for his advice about this section of this book.

   Return to note reference.

5. For more on the postal system as an information network, see Henkin (2006).

   Return to note reference.

6. The Viral Texts project corpora do include some titles published earlier than 1840, but the coverage in the corpus becomes much more substantial around the middle of the century. This is due to the expansion of newspapers during the period, as we will describe below, but also because most mass digitization efforts started with papers from around 1900 and have gradually digitized further in the past. So few newspapers from the 1800s-1830s have been digitized that, while we will sometimes reference particular texts reprinted during those earlier periods, it is difficult to meaningfully compare trends at scale between those decades and later ones. Thus this chart starts in 1840.

   Return to note reference.

7. Thinking about historical parallels, we cannot help but link “the upper ten thousand” or “upper ten” from the nineteenth century with “the one percent” from the twenty-first. In both cases, the descriptive phrase about an economic subset of the population becomes a shorthand for critiques of wealth inequality and exploitation. That shorthand is doing significant work, we would argue, in the Sentinel’s sly critique of the Mirror in this selection.

   Return to note reference.

8. The “inst.” in this article is a common newspaper abbreviation for “instant,” which designates the current month. Since this article was printed in the Cadiz Sentinel on February 19, 1845, we would assume the reprinted article appeared in the New York Evening Mirror on February 10th, 1845.

   Return to note reference.

References

1. Abebe, Rediet, Solon Barocas, Jon Kleinberg, Karen Levy, Manish Raghavan, and David G. Robinson. 2020. “Roles for Computing in Social Change.” arXiv:1912.04883 [Cs], January. https://doi.org/10.1145/3351095.3372871.

2. Algee-Hewitt, Mark, Sarah Allison, Marissa Gemma, Ryan Heuser, Franco Moretti, and Hannah Walser. 2016. Canon/Archive. Large-scale Dynamics in the Literary Field. Stanford Literary Lab, Pamphlet 11.

3. Casey, Jim, and Sarah H Salter. 2020. “Challenges and Opportunities in Editorship Studies.” American Periodicals 30 (2): 5.

4. Cordell, Ryan. 2017. ““Q i-Jtb the Raven”: Taking Dirty OCR Seriously.” Book History 20: 188–225. https://doi.org/10.1353/bh.2017.0006.

5. Fagan, Benjamin. 2016. “Chronicling White America.” American Periodicals: A Journal of History & Criticism 26 (1): 10–13.

6. Fyfe, Paul. 2016. “An Archaeology of Victorian Newspapers.” Victorian Periodicals Review 49 (4): 546–77. https://doi.org/10.1353/vpr.2016.0039.

7. Garvey, Ellen Gruber. 2012. Writing with Scissors: American Scrapbooks from the Civil War to the Harlem Renaissance. Oxford University Press.

8. Gitelman, Lisa, and Thomas S. Mullaney. 2021. “Nineteenth-Century Media Technologies.” In Information: A Historical Companion, edited by Ann Blair, Paul Duguid, Anja-Silvia Goeing, and Anthony Grafton. Princeton: Princeton University Press.

9. Hayles, N. Katherine. 2002. Writing Machines. Cambridge, Massachusetts: MIT Press.

10. Henkin, David M. 2006. The Postal Age: The Emergence of Modern Communications in Nineteenth-Century America. University of Chicago Press.

11. Lee, Benjamin. 2021. “Compounded Mediation: A Data Archaeology of the Newspaper Navigator Dataset.” Digital Humanities Quarterly 015 (4).

12. Libraries, Penn State University, and National Endowment for the Humanities. n.d. “Clearfield Republican. [Volume].” Accessed May 11, 2021.

13. Mak, Bonnie. 2014. “Archaeology of a Digitization.” Journal of the Association for Information Science and Technology 65 (8): 1515–26. https://doi.org/10.1002/asi.23061.

14. Milligan, Ian. 2013. “Illusionary Order: Online Databases, Optical Character Recognition, and Canadian History, 1997.” The Canadian Historical Review 94 (4): 540–69.

15. Nord, David Paul. 2006. Communities of Journalism: A History of American Newspapers and Their Readers. Urbana: University of Illinois Press.

16. Pettegree, Andrew. 2021. “Newspapers.” In Information: A Historical Companion, edited by Ann Blair, Paul Duguid, Anja-Silvia Goeing, and Anthony Grafton. Princeton: Princeton University Press.

17. R. Hoe & Company. 1873. [Catalogue of] Printing Machines . . . Hand-Presses, Self-Inking Machines, Etc. New York.

18. Schmidt, Benjamin M. 2016. “Do Digital Humanists Need to Understand Algorithms?” In Debates in the Digital Humanities.

19. Senchyne, Jonathan. 2017. “Paper Nationalism: Material Textuality and Communal Affiliation in Early America.” Book History 19 (1): 66–85.

20. Senchyne, Jonathan. 2019. The Intimacy of Paper in Early and Nineteenth-Century American Literature. Amherst: University of Massachusetts Press.

21. Swoope, Roland D. (Roland Davis). 1911. Twentieth Century History of Clearfield County, Pennsylvania, and Representative Citizens. Chicago, Ill., Richmond-Arnold publishing co.

22. Trott, Louisa. 2022. “Documenting Lost African-American Newspapers.” The National Endowment for the Humanities.

23. Tucher, Andie. 2010. “Newspapers and Periodicals.” In An Extensive Republic: Print, Culture, and Society in the New Nation, 1790–1840, edited by Robert A. Gross and Mary Kelley. UNC Press Books.

24. University of Tennessee, and National Endowment for the Humanities. n.d. “Brownlow’s Knoxville Whig. [Volume].” Accessed May 11, 2021.

25. Weeks, Lyman Horace. 1969. A History of Paper-Manufacturing in the United States, 1690–1916. [New York] B. Franklin.