When theory and practical know-how began to reinforce each other

A two-way explanation for growth

Economic historian Joel Mokyr argues that modern growth emerged partly because two kinds of knowledge began reinforcing one another. Propositional knowledge describes how nature works, including theories and empirical observations. Prescriptive knowledge describes how to manipulate it, from building machines to navigating ships.

The hypothesis requires more than science supplying discoveries to inventors. Practical experience must also feed back into understanding: an instrument reveals an unexpected phenomenon, or an engineering experiment exposes a theory’s weakness. Both directions must become productive for the loop to be positive.

Julius Koschnick tests this pattern in an arXiv preprint, supplied here as version 3, examining Britain before and during the Industrial Revolution. The supplied record does not confirm journal publication or peer review. He presents the study as the first quantitative test of Mokyr’s feedback-loop hypothesis.

Reading the historical knowledge economy

The source material includes 285,985 unique publication titles from the English Short Title Catalogue for 1600–1800, alongside historical patent descriptions. These titles were often miniature abstracts, averaging about 55 words rather than resembling today’s brief book titles.

The central analysis uses selected scientific and technical fields, not the entire catalogue. Theoretical fields include applied physics, astronomy, mathematics, chemistry and encyclopedias. Practical fields include technical instructions for trades and agriculture, navigation, scientific instruments and patents. Life sciences are excluded under the paper’s focus on mechanically oriented innovation.

Koschnick uses SteamBERTh, a language model adapted to historical scientific and technical texts, to compare their meanings. His innovation index asks whether a publication resembles its field’s future more than its past. The baseline compares it with the 20 most similar earlier titles and the 20 most similar titles published during the following 20 years.

A separate index measures received knowledge spillovers. A practical title, for example, scores higher if it resembles recent theoretical publications more closely than comparable earlier practical titles did. This is a text-based proxy for drawing on another field, not a direct record of someone reading or applying a particular work.

The paper checks these measures against historical patent citations, book editions and authors publishing across fields, alongside modern citation and patent-value data. Positive relationships support their usefulness, although raw historical correlations are modest.

A change around the mid-eighteenth century

The main question is whether publications drawing more heavily on the other knowledge sphere were also more innovative, after accounting for subject, publication year and title length.

Koschnick reports a negative relationship between theoretical spillovers and practical innovation before 1720. The relationship was approximately neutral in 1720–1739 and positive in 1740–1759 and 1760–1779. In the reverse direction, practical spillovers were broadly unrelated to theoretical innovation before 1740, then positively associated with it.

For 1760–1779, the paper reports that doubling the received-spillover index from theoretical knowledge was associated with 10.6% higher practical innovativeness. Doubling the reverse-direction index was associated with 8.3% higher theoretical innovativeness. These percentages describe changes in constructed indices, not economic growth rates.

Patent-only results show a positive, statistically significant association with theoretical spillovers in 1760–1779. Patent-citation results show a similar late-century pattern. This extends the evidence beyond books, but does not directly estimate effects on national output.

Connections and methods mattered

During 1740–1779, Royal Society fellows and Society of Arts members had higher measured spillovers. University education and academic careers were particularly relevant to incorporating theoretical knowledge into practical publications. Koschnick interprets these associations as evidence for the importance of expertise and knowledge-sharing networks.

Method also appears relevant. Textual similarity to precise measurement strengthened the relationship between theoretical spillovers and practical innovation. In the reverse direction, Newtonian mechanics, precise measurement and the scientific method showed positive interactions when examined separately; only Newtonian mechanics remained statistically significant when all three were included together.

Support, not causal proof

Koschnick tests whether changing vocabulary, writing style or publication volume could explain the findings. Comparisons with other subject fields, fixed-size comparison pools, shuffled publication years and simplified rewritings generally support the central pattern. Not every specification is equally decisive: including patents as spillover sources makes the final-period reverse-direction coefficient statistically insignificant. The baseline excludes them because access to patent specifications was costly and uneven.

The interpretation requires textual convergence to reflect knowledge use, rather than vocabulary alone, and no unaccounted-for common shock to raise both indices within a field and year. The author explicitly describes the estimates as associations, not causal effects.

Other boundaries remain: secret inventions are absent, and semantic similarity cannot reliably distinguish a successful argument from a rebuttal discussing the same topic. The contribution is therefore specific: evidence consistent with theory and practice becoming mutually productive by the mid-eighteenth century, rather than proof that this feedback alone created modern growth.