We’re quite familiar today with the idea of news cycles: the fluctuating window of time it takes for a full turnover of major news items. Underlying these variable, context-driven cycles are other, structural patterns of news making and circulating. These are what I like to call news rhythms.
The difference is partly a classic superstructure / base distinction in that many of the factors contributing to news rhythms are economic or industrial. Historically, the newspaper was as much a literary and social medium as an informational one though, so I don’t mean “news” in the strict sense. News rhythms are also shaped by cultural or even arbitrary factors, as when some TV timeslots are shaped by the after-effects of a particularly popular program. To some degree they also take on a perpetuating momentum of their own; established patterns sometimes redouble “merely” by the force of their own precedent. Hence, in part, “rhythms.”
Using data from Chronicling America, I’ve been exploring a few ways to represent these underlying rhythms in nineteenth- and early twentieth-century newspapers in order to better understand how they impact the results of text mining research.
Figure 1 shows the distribution of pages printed on each day of the week throughout the years for 1867, 1889, and 1918. Because the contents of Chronicling America reflect the digitization choices of institutions rather than the true number of pages printed, I have represented totals as a proportion rather than a raw count: the number of pages for each day divided by the number of available pages for that year in Chronicling America. This makes for a more stable metric of comparison across years.1 The black line corresponds to the second y axis, at right, representing the average number of pages per issue.
Historically speaking, Figure 1 graphs the growth of the Friday issue (as the last work day on which to obtain a paper before the weekend) and the emergence of the Sunday edition (an aberration prior to the end of the 1860s). Some of these papers were weeklies; some were the weekly editions of dailies; some were dailies, which themselves fluctuated in length by day of week. As the jump in page count suggests, this quickly became its own distinctive entity. It was often in fact billed as a “Sunday magazine,” which naturally did not thrill the weekly magazines that had to start competing with them.
But again, there’s an element of arbitrariness to news rhythms. In the nineteenth century the day of publication for weekly newspapers and magazines was often something of a local tradition – or a reaction to it on the part of a competitor seeking more open ground. The preference for Wednesdays vs. Thursdays (or, later, Thursdays vs. Fridays) sometimes depended on place.
Representing page count over time provides a sense of the pattern and regularity underlying this distribution. Figure 2 uses the same data with the x axis is grouped by date instead of day. As for the y axis, I wanted to capture a sense of the true difference in total volume of newsprint from year to year, since the number of newspapers and pages per issue both significantly increased. So I used a proportion again (pages per date divided by yearly total) multiplied by the total number of newspapers published in that year.2
The change in the distribution of pages published on each day in Figure 1 is represented here in a sharpening of peaks from 1867 to 1889 to 1918. With an increase in the number of pages per day came an increase in variability from day to day as well. As the number of newspapers almost tripled from 1867 (5871) to 1889 (16279), the amount of variability almost doubled; as the number of newspapers almost quadrupled from 1867 to 1918 (22842), the mount of variability almost tripled. The end-of-year funkiness in all three corresponds with Christmas.
I find this visualization beautiful. The wave appearance should make clear an additional reason why I like the phrase “news rhythms,” and it reflects a vital experiential element of print. This manner of visualization highlights newspapers’ contributions to the tempo of daily life as print news became ever more ubiquitous and indispensable over the nineteenth century, not only in print-glutted world of the urban daily but in the territory of the country weekly, whose denizens increasingly took multiple papers and exchanged copies with friends and neighbors.3
Like sonic waves, of course, news rhythms travel over both time and space: that is, the time of transmission or delivery. Still, many readers – especially before the maturation of Rural Free Delivery over the first decade of the 1900s – retrieved their newspapers at post offices, general stores, news stands, or their printers and planned their days or weeks accordingly. The readerly experience of news rhythms was thus contingent on a variety of factors that could cause either a lag (reader location, especially if rural) or stagger (receiving additional papers from distant locations) but not necessarily.
I enjoy quibbles, so I’ll note that this isn’t a completely stable metric of comparison: it is still possible that, by chance, all of the state-level selection committees picked a grossly disproportionate number of, say, Thursday-publishing papers from 1867 relative to the true number of Thursday-publishing papers from 1867. Because each of these years of data encompasses over 200 papers, because Figure 2 attests to the consistency of these figures, and because both align with other forms of evidence, though, I believe they are at least generally reliable. ↩︎
For 1889 and 1918 I used N. W. Ayer & Son’s Newspaper Annual. For 1867 I fibbed a bit and used the 1870 data, the closest I had on hand, from S. N. D. North’s rich 1880 census report, History and Present Condition of the Newspaper and Periodical Press of the United States. ↩︎
David Henkin wrote an nice article fairly recently on the “hegemony of the weekly regime” in the nineteenth century that draws considerably on newspapers. David Henkin, “Hebdomadal Form: Diaries, News, and the Shape of the Modern Week.” Representations 131.1 (2015): 52-67. www.jstor.org/stable/10.1525/rep.2015.131.1.52 ↩︎