How has the average length of a scientific article changed over time? The answer depends on the purpose of the ‘article’ (a letter, a preliminary announcement, a fully-detailed research monograph); the space available in the printed journals; and the contemporary fashion for scholarly writing. But here, nonetheless, are some insights from the history of Royal Society publishing. [NB see caveats below]
As the Society’s oldest, longest-running journal, the Transactions gives us the longest view: the length of published items was growing through the 18th and 19th centuries (which helps to explain why publishing costs were so expensive by 1890); falling in the early 20th century (perhaps under pressure from treasurers, but note that there was also a more general shift to using shorter article lengths in more-rapid publication venues – including the Proceedings); and falling most notably around 1970. That final drop looks like a change in formal page limits, though we’ve found no direct evidence. Historically, the Transactions didn’t have page limits (unlike the Proceedings), other than a (very notional) 40 pages.
This second graph shows the different practices between the Society’s two varieties of journal in the 20th century – but less difference in practice between the physical sciences (series A) and life sciences (series B) than one might have expected.
After 1905, the Proceedings specialised in shorter articles (published approximately monthly), while the Transactions remained a venue that would take much longer articles. The majority of Royal Society articles were published in the Proceedings throughout this period. There was a notional limit of 12 pages on Proceedings articles early in the century, later rising to 24 pages. When the journals were relaunched in 1990, Proceedings B introduced a strict page limit – as is very clear in the figure!
Warnings and limitations
The data above are taken from a spreadsheet that is maintained by the Society’s publishing staff. That spreadsheet includes the number of pages per issue, and number of articles per issue, of all RS journals from 1665 to present (but without distinguishing between text or images). With help from the St Andrews research computing team, that data has been turned into counts per year (rather than per issue or per volume) totals, for ease of comparison.
The page counts make no allowance for the occasional change of paper size, margins or text size, nor for the quantity of illustrations.
Average article length was obtained by dividing the number of pages by the number of articles per year. For the reasons stated above, it is a rough approximation, rather than an exact measure, but it does a useful job of giving us a sense of the trends over time.