How fast did news travel?

On the evening of 14 April 1865 Abraham Lincoln was shot at Ford's Theatre. By the next morning the news was in print in Cleveland, Nashville and New York, and the same day in Gold Hill, Nevada, 3,643 km away. But in Texas it took weeks. How fast did news really travel, and what slowed it down? I took four great American stories, from the telegraph age to the radio age, and checked 13,233 digitised issues of 936 newspapers for the first day each paper printed the news.

The four are Lincoln's assassination (1865), Custer's defeat at the Little Bighorn (1876), the sinking of the Titanic (1912) and the attack on Pearl Harbor (1941). The pages come from Chronicling America, the Library of Congress's archive of digitised newspapers, as turned into searchable text by the American Stories project at Harvard. The map above plays each story day by day.

01 · The answerThe telegraph had already won by 1865

The surprise is how little changed. A daily paper that printed Lincoln's death waited a median of 1 day; so did a daily printing the Titanic in 1912 (1) and Pearl Harbor in 1941 (1). Once a story reached a telegraph office, it was in every connected city by the next edition, whether that city was in Ohio or Nevada. What took time was the last step: most American papers were weeklies, and a weekly printed the news when its press day came round. Weeklies took a median of 7 days for Lincoln, 4 for the Titanic and 5 for Pearl Harbor.

Median days from the event to the first report, for daily papers (filled) and weekly papers (open).

So for most papers, the news missed no issue at all: 62% of the papers that reported Lincoln's death did so in their first issue after it, and 65% for the Titanic. The curves below climb in steps, each step a press day.

Share of the papers with an issue in the first fortnight that had printed the story, by days after the event.

02 · 1865A country split by war

Lincoln died as the Civil War ended, and much of the South's telegraph and railway was wrecked. Papers in the states of the former Confederacy took a median of 11 days to print his death, against 6 days elsewhere. In South Carolina, where Sherman's army had passed weeks before, it was 10.5 days. The Texas papers in the sample took three to four weeks, and in Honolulu, then capital of an independent kingdom reached only by ship, The Pacific Commercial Advertiser printed it after 29 days. Yet Gold Hill in Nevada, on the transcontinental line, had it the day after the shooting.

Each dot is a newspaper's first report of Lincoln's death; red marks the states of the former Confederacy.

03 · 1876Custer: the slowest mile

The Little Bighorn shows where the real delay lay. The battle was fought on 25 June 1876 in Montana Territory, days from the nearest telegraph. No paper in the sample printed it for 11 days. Then, on 6 July, it was everywhere at once: 12 papers in 9 states and territories printed it that day, from New York and Washington to Cheyenne and Nevada. Distance from the battlefield made no difference after that; the wire did not care. The Montana weeklies in the sample show it late, but their surviving pages are few, so that says more about the archive than about Montana.

Newspapers printing Custer's defeat for the first time, by date and region.

04 · DistanceFar did not mean late

Plot each paper's delay against its distance from where the news entered the wires and there is almost no pattern among the dailies. In 1912 the two Honolulu dailies printed the Titanic news on 15 April, the day she sank, 7,982 km from New York; in 1941 the Daily Alaska Empire in Juneau had Pearl Harbor the same day. The spread in each panel comes from how often the paper was printed, not where it was.

Days to the first report against great-circle distance from Washington (1865, 1941), the battlefield (1876) or New York (1912).

05 · MethodHow I did this

For each event the dataset holds every digitised page in a window of six to ten weeks afterwards, with a flag where the page text matched the event's name (Lincoln, Custer, Titanic, Pearl Harbor) and a short snippet round the match. A name is not a story: Lincoln County and Custer City were always in the news. So a page counts as reporting the event only if it is dated on or after the first day the news could have been printed and its snippet contains a word about the event itself (assassination, massacre, iceberg, attack, and so on). I read the first report of every paper that was early, short of a clear word, or might be about something else (290 snippets) and threw out 9; in a random check of 60 others, 59 were plainly about the event and 1 was unclear.

For each paper I then took the first issue after the news could have arrived and the first issue that reports it, counting days from the event and the issues in between. A paper counts if it has a digitised issue in the first fortnight (145 papers for Lincoln, 216 for Custer, 566 for the Titanic, 69 for Pearl Harbor). A paper came out daily if its issues were at most two days apart, weekly if five or more. Each paper is placed at the town in its Library of Congress title, matched to Wikidata and GeoNames (850 of 936 placed).