Meta: no revolution / gradualism

Thesis

There was no “Industrial Revolution.” The phrase, coined by Arnold Toynbee in the 1880s and popularized by T.S. Ashton’s mid-20th-century textbook, imposes a revolutionary discontinuity on what was in fact a century-long gradual transformation. When you reconstruct the quantitative record carefully:

British aggregate growth was modest before 1830 — perhaps 0.3–0.5% per capita per year.
Most of the population was not employed in the “revolutionary” sectors (cotton, iron, steam); agriculture, services, and traditional crafts still dominated.
The sharp take-off, if it happened at all, happened in the 1830s–1850s with railways and the factory system, not in the 1770s–1790s with Arkwright and Watt.
Real wages for workers were flat or falling through the early IR; stature declined; the “revolution” did not obviously improve ordinary lives until the mid-19th century.

At the extreme (Crafts, Harley), the quantitative revisionists argue for replacing “revolution” with a gradual transition framing. A parallel current (de Vries) argues the real behavioral transformation — the “industrious revolution” — happened a century earlier, as households reorganized around market labor and consumption. Together, these arguments destabilize the IR as a well-defined object of explanation.

This is a meta-position: it doesn’t compete with the other positions on their own terms; it argues that the question those positions are answering is framed wrong.

Lead proponents

Nicholas Crafts (with C. Knick Harley) — the quantitative revisionist program. British Economic Growth during the Industrial Revolution (1985) and the 1992 Crafts–Harley restatement produced the revised aggregate growth estimates — roughly 0.3–0.5% per-capita growth through 1760–1830 vs. the older Deane–Cole figures — that destabilized the “revolutionary” framing. The framework being superseded was Rostow’s 1960 “take-off” model, which dated the British take-off to 1783–1802 on the basis of an assumed 10% investment-rate threshold that the quantitative work did not support. Older textbooks (Rostow; Landes 1969) still treat the IR as a discontinuous take-off; the Crafts revision has replaced this in professional economic history but popular and cross-disciplinary treatments often lag.
Jan de Vries — The Industrious Revolution (2008). Shifts the locus of transformation a century earlier, arguing that the real household-level change — more market labor, more consumer demand, more engagement of women and children in wage work — preceded and enabled the later technological story.
Stephen Broadberry (with Bruce Campbell, Alexander Klein, Mark Overton, and Bas van Leeuwen) — British Economic Growth 1270–1870 (2015) extends the national accounts back six centuries from the output side. The reconstruction shows English/British per-capita GDP growing slowly but persistently from the late Middle Ages, with no sharp late-18th-century kink — supporting the gradualist reading that the IR is the steepening of a long incline, not the lighting of a fuse. It also relocates much of the “rise” to earlier episodes (the post-Black-Death rise, the 17th-century commercial expansion), further deflating the 1760–1830 window.

The Crafts–Harley deflation

The quantitative core of the position is a specific downward revision, and it is worth being precise about what changed. The mid-century orthodoxy was the Deane & Cole national accounts (British Economic Growth 1688–1959, 1962), which — combined with Rostow’s “take-off” model — implied British industrial output growing at ~3–4% a year and per-capita income accelerating sharply from the 1780s. Crafts (1985) rebuilt the output index sector by sector and found that Deane & Cole had over-weighted the fast-growing modern sectors (cotton, iron) and mis-measured the large slow-growing ones (agriculture, services, traditional manufactures). The Crafts–Harley (1992) restatement, responding to challenges, settled the revised series: industrial output growth of perhaps ~2% rather than ~3–4%, and real GDP per capita growth of only ~0.2% in 1760–1801 and ~0.5% in 1801–31, accelerating to ~1% only after 1830. The “revolution,” on these numbers, is a 19th-century phenomenon arriving a half-century after the canonical Arkwright–Watt opening.

The methodological point that gives the revision its force is an aggregation argument: a sector growing at 6–7% a year (cotton) contributes little to aggregate growth while it is only 4–5% of the economy, and the aggregate number is dominated by the 35%+ of the economy that was still agriculture growing near 0.5%. The dispute with the rehabilitation position is precisely whether that aggregation logic is the right lens or a way of hiding the engine inside the average.

Key arguments

Aggregate growth before 1830 was slow. Crafts–Harley’s revised real-GDP-per-capita series shows British growth of ~0.3–0.5%/year in 1760–1830, accelerating to ~1.2–1.5%/year only in 1830–1870. By modern standards this is a gentle trend, not a revolution.
Cotton and iron were small shares of the economy. Cotton at peak employment (~1830) was ~4–5% of the British labor force. Iron and coal smaller. Agriculture in 1800 was still ~35% of employment. The “revolutionary” sectors were small relative to the whole economy for most of the “revolutionary” period.
The industrious revolution predates the IR. Dutch, British, and Flemish household data from the 16th–17th centuries show rising market-labor participation, expanding consumer demand for “new” goods (sugar, tea, cotton, imported manufactures), and a shift in time allocation away from leisure and household self-production toward wage labor. This transformation is much larger in its population impact than the late-18th-century technological changes.
The “take-off” periodization doesn’t fit the aggregate data. Rostow’s 1960s-vintage “take-off” model of 1783 or thereabouts was always quantitatively weak, and the revised GDP series kill it outright. If there was a take-off, it was in the 1840s–50s with railways — long after the conventional IR dating.
Real wages and welfare tell a different story. Feinstein’s wage series (1998) and the anthropometric literature (Komlos) show flat-to-falling real wages and declining adult male stature through ~1820–1830. The “revolution” did not, for most workers, materialize as improved living standards until the second half of the 19th century. If the payoff is this delayed, calling the early period “the payoff-generating event” is strange.

The word “revolution” as a Victorian artifact

Part of the position is historiographical rather than quantitative: the very concept of an “Industrial Revolution” is a retrospective construction whose connotations the data do not support. The phrase entered English through Arnold Toynbee’s posthumously published Lectures on the Industrial Revolution in England (1884), which framed the period 1760–1840 as a sudden, catastrophic rupture — a moral and social calamity as much as an economic one — and the framing was hardened into textbook orthodoxy by T. S. Ashton’s The Industrial Revolution 1760–1830 (1948) and given a dramatic growth-theoretic shape by Rostow’s “take-off into self-sustained growth” (1960), which dated the British take-off to 1783–1802 on the strength of an assumed jump in the net-investment rate (from ~5% to ~10% of national income) that the later national-accounts work could not find. The gradualist point is that “revolution” smuggles in a discontinuity — a single generation’s decisive break — that the reconstructed series simply do not contain. What the data show is a slow steepening; the revolutionary language belongs to the polemics of the 1880s, not to the economy of the 1780s.

The welfare and anthropometric strand

A distinct and more radical line questions whether anything improved for ordinary people in the supposed revolutionary decades. Charles Feinstein’s authoritative real-wage reconstruction (1998) found that working-class real earnings were broadly flat from the 1780s to the 1820s and rose decisively only from the 1840s — a half-century “pause” in living standards across the canonical IR window. The anthropometric literature (John Komlos and others) reinforces this: average adult male stature, a proxy for net nutritional status in childhood, appears to have declined for cohorts born in the early industrial decades, recovering only mid-century. The “no-take-off” inference is that from the worker’s-eye view there was no obvious break at all, and that calling a period of stagnant wages and shrinking bodies “the payoff-generating event” of modern growth is, at minimum, an odd use of language. This strand sits in tension with the rehabilitation position, which accepts the welfare data but insists the structural transformation was real regardless of its short-run welfare cost.

The long-run reconstruction: growth was old

Broadberry and collaborators’ British Economic Growth 1270–1870 (2015) is the gradualist case at its longest reach, and its findings reframe the question. Building English GDP from the output side back to the late thirteenth century, the reconstruction shows that per-capita income did not sit flat through the medieval and early-modern centuries waiting for an 1780s explosion. Instead it rose substantially after the Black Death (the post-1348 collapse in population raised land and labour per head and pushed measured per-capita output up sharply), gave much of that back during the sixteenth-century population recovery, and then resumed a slow, irregular climb across the seventeenth and eighteenth centuries — so that England had already pulled meaningfully ahead of the European pack well before the canonical IR. The series shows no single sharp kink at 1760 or 1780; the late-eighteenth-century acceleration is a steepening of an incline already centuries long. The gradualist reading is that the IR is the visible final phase of a slow escape from the Malthusian band, not the moment of escape itself — which both deflates the “revolution” and pushes the explanatory burden back onto the much earlier divergence the long series reveals (and toward the Great Divergence debate over when, exactly, northwestern Europe began pulling away).

The industrious revolution

The gradualist current has a second, quite different limb, owed to de Vries in The Industrious Revolution (2008). Where Crafts–Harley deflate the technological event quantitatively, de Vries relocates the transformative event a century earlier and into the household. His claim is that from roughly the mid-17th century, Northwestern European households reallocated their time: they worked more days and more hours in the market, drew women and children further into wage and proto-industrial labour, and reduced leisure and self-provisioning — not because they were forced to but in order to buy a widening basket of new market goods (sugar, tea, coffee, tobacco, cottons, crockery, clocks). This is a demand-side, behavioural transformation that precedes and enables the supply-side technological one, and it shows up in measures the technology-centred story ignores: rising consumer-durable ownership in probate inventories, falling self-sufficiency, the spread of the “respectable” consumption package down the social scale.

The industrious-revolution thesis is gradualist in the specific sense that it denies the late-18th-century technological cluster the status of the turning point — the real change in how ordinary people organized work and wanted goods was already a century old by the time the spinning jenny appeared. De Vries himself, importantly, treats the behavioural and technological changes as complements in one long transformation, which is why this limb sits uneasily with the harder “no revolution at all” reading: it relocates the revolution rather than denying it.

Key evidence

Crafts–Harley revised GDP series (1992 and updates) — the standard modern aggregate series, substantially below Deane–Cole’s mid-20th-century estimates.
Broadberry et al. British GDP 1270–1870 (dataset) — the even longer-run reconstruction showing growth was an old phenomenon, not a late-18th-century break.
Sectoral employment shares — Wrigley, Shaw-Taylor, and the Cambridge Group’s occupational-structure work.
De Vries’s household-level reconstructions — consumption patterns, female and child labor participation, time-use evidence for 16th–18th-century NW Europe.
Real-wage series (Feinstein, Allen, Clark) — all broadly agreeing that workers’ conditions did not improve, and likely worsened, through 1780–1830.

Major critiques

— BERG & HUDSON: aggregates hide the structural revolution. Even if aggregate growth was modest, the composition of the economy was transforming — cotton growing at 6–7% a year, iron output rising roughly twenty-fold across 1760–1830, the factory displacing the workshop. Berg & Hudson (1992) made this case formally in the same 1992 Economic History Review volume as the Crafts–Harley restatement, arguing that aggregate per-capita GDP is simply the wrong instrument for detecting a sectoral and organizational revolution. The rehabilitation position is the polar opposite of this one and forms the meta-debate within the meta-debate.

— The second-IR discontinuity: even granting a gradual first IR, something discontinuous unmistakably happened by the later 19th century — sustained per-capita growth became the new worldwide normal and the Malthusian regime was permanently broken. Something caused that, and a story that smooths the whole period into a gentle trend struggles to explain why the trend did not, as every previous efflorescence had, peter out.

— The industrious revolution is complementary, not rival: de Vries himself frames the behavioural change as enabling the technological one, parts of a single long transformation. Treating “industrious” against “industrial” as competing periodizations overstates a tension its own author denies, and concedes that there was a transformation to locate.

— Real-wage stagnation is predicted, not anomalous: during rapid structural change with high investment and fast population growth, aggregate output can rise while per-worker consumption lags — the Malthusian-relief lag is exactly what standard growth models predict. The flat early-IR wage series is therefore weak evidence against a real transformation, only against an immediate welfare payoff.

— “No revolution” leaves the central fact unexplained: sustained growth did escape the Malthusian world, and it did so first in Britain. A weak quantitative break is still a break demanding explanation. The harder gradualist rhetoric sometimes reads as though the question itself were illegitimate — but the question is the deepest one in economic history, and deflating the magnitude of the answer does not dissolve it. This is why most synthesizers (Mokyr, Allen, Broadberry) absorb the gradualist numbers and then proceed to explain the transition anyway.

Status

Contested, and complicated. The empirical core — that the quantitative IR was slower and smaller than mid-20th-century textbooks suggested — is broadly accepted and has reshaped the debate. But the strong rhetorical claim that there was “no revolution” is much less widely accepted. Most modern synthesizers (Mokyr, Allen, Broadberry) concede the gradualist numbers and adjust their explananda accordingly — explaining the transition to modern growth, rather than a sudden discontinuity. The meta-critique is now absorbed by the literature rather than dismissed, which is a particular kind of victory for the gradualist program.