Rehabilitating the Industrial Revolution

Thesis

The Crafts–Harley quantitative revisionism (Crafts 1985; Crafts & Harley 1992) undercut the old “take-off” image of the Industrial Revolution by showing that aggregate per-capita GDP growth was slower than the textbooks claimed. But this is the wrong lens for the question. Aggregate GDP growth is a poor summary of what was actually transformative about the late-18th-century British economy: the emergence of factory production, the wholesale restructuring of sectoral composition toward cotton and iron, the reorganization of regional specialization, the transformation of women’s and children’s labour, the collapse and re-formation of demographic patterns, and the political-class formation that accompanied all of this. Each of these is a genuine revolution in the meaningful sense. Modest aggregate growth was a by-product of the fact that slow-growing agriculture and services were large shares of the economy — but the engine of modern industrial society was nonetheless being assembled, sector by sector, in exactly the period the aggregate numbers seem to dismiss.

The Berg and Hudson (1992) paper that gives this position its name is framed explicitly as a response to Crafts–Harley: “rehabilitating” the IR from the quantitative revisionists’ deflation. The argument is that both economic and social historians have been too easily persuaded to downgrade the IR because of aggregate numbers that systematically understate the things that matter.

This position competes directly with the gradualism / no-revolution meta-position; the two are the polar views on what the IR actually was, and form the meta-debate within the meta-debate.

Lead proponents

Maxine Berg — British economic historian at Warwick; long-running work on the material culture, regional organization, and consumer dimensions of the IR. With Hudson, the 1992 “Rehabilitating” paper.
Pat Hudson — British economic historian at Cardiff; work on regional industrialization (West Riding), finance, and women’s labour in the IR. Co-author of the 1992 paper.
Jane Humphries — independently in this tradition, though with her own research program; Childhood and Child Labour (2010) and the spinning-wages work (2019) document the qualitative transformations in work and labour markets that Berg and Hudson point to.
A broader “social-historical” tradition — E.P. Thompson’s The Making of the English Working Class (1963), Raphael Samuel (whose “Workshop of the World,” 1977, argued that hand technology and sweated labour, not just steam-driven factories, expanded massively in the IR), and the History Workshop tradition — is the deeper root of this view, even if later Berg–Hudson-style scholars have moved the ground to quantitative-social synthesis.

The structure of the Berg–Hudson argument

Berg & Hudson’s 1992 intervention is best read as a methodological brief against the Crafts–Harley aggregate, not a denial of its arithmetic. Their case has three moves. First, the aggregate growth rate is a weighted average dominated by the large, slow sectors (agriculture, services, traditional crafts), so it is structurally incapable of registering a revolution confined at first to a few small sectors — the measure is mis-specified for the question. Second, the national accounts rest on output and productivity indices that systematically undercount exactly the labour the IR mobilized: women, children, and the proto-industrial and sweated hand-trades that expanded alongside the factories. Because the wage and output series are built around adult-male, formal-sector work, they miss the demographic majority of the industrial workforce and understate both output and the intensification of labour. Third, the things that make the period a watershed — the factory as a form of work discipline, the spatial concentration of industry, the proletarianization of labour — are qualitative structural changes that a per-capita GDP figure was never designed to capture and whose absence from the index says nothing about their reality.

The polemical edge is that economic history, in Berg and Hudson’s telling, allowed a measurement convenience (GDP per head) to redefine its object, “deflating” the IR because the chosen instrument could not see it. The gradualist reply — that an economy in which transformative sectors are small must show modest aggregate growth, which is a description and not an oversight — is the crux of the meta-debate.

Key arguments

Aggregate GDP is the wrong measure. Cotton grew at 6–7% per year across the 1780s–1830s; raw-cotton imports rose from ~2.5M lb in 1760 to ~592M lb by 1840. British pig-iron output rose from roughly 25–30,000 tons around 1720 to ~250,000 tons by 1800 to ~2.7M tons by 1852 — a hundred-fold increase. Coal output rose from ~3M tons/year in 1700 to ~15M by 1800. These sectoral transformations are historically unprecedented; no prior episode of human economic history shows comparable sustained sectoral growth rates. The aggregate number is low only because these sectors remained small shares of a mostly-agricultural economy until mid-century, when the composition finally followed and aggregate growth accelerated. Berg–Hudson’s point: you cannot judge a sectoral revolution by an aggregate per-capita average.
The factory system was a genuine discontinuity in the organization of production. For most of human history, production was organized around households or small workshops. The factory — a centralized, capital-intensive, wage-labour site of coordinated mass production — was a new thing, emerged first in British cotton spinning in the 1770s–1790s, and by 1850 was the dominant mode of production for industrial goods. This is a revolution in the organization of work, regardless of what aggregate GDP did.
Women’s and children’s labour was transformed. Humphries’s work on apprenticeship and child labour, and the Humphries–Schneider work on spinning, document how the IR transformed the working conditions of women and children in ways that mattered enormously for social and demographic outcomes — and that the male-adult-wage aggregate series systematically miss. Any account that focuses on adult male builder wages in London (as much of the Allen-tradition data does) will by definition miss the largest demographic experience of the IR’s workforce.
Regional specialization intensified dramatically. Lancashire (cotton), the West Riding (woollens), South Wales and Staffordshire (iron and coal), the West Midlands (metalware), the Potteries, Newcastle (coal) — the spatial concentration of specific industries in specific regions, at scales unknown before the IR, is a structural transformation of British economic geography. Aggregate national-level growth numbers hide this entirely.
Demographic transformation accompanied the economic one. Population roughly doubled in the IR century; urbanization rates rose from ~25% to ~50% between 1780 and 1850; the peculiar Northwestern European Marriage Pattern began to loosen. These are fundamental structural changes in how British society was organized and reproduced. Aggregate per-capita GDP growth at 0.3% per year tells you nothing about this.
Political-class formation was a consequence, not a sideshow. The emergence of a self-conscious industrial capitalist class, a factory proletariat, a reforming middle class with Parliamentary ambitions — the whole political structure of Victorian Britain, culminating in the Reform Acts and the eventual democratization of the state — was a direct consequence of the economic reorganization the quantitative revisionists downplay.

The factory as a revolution in work

The qualitative claim that does the most work for the position concerns the organization of production, and it has its own deep literature. For nearly all of human history goods were made in households and small workshops, by people who largely controlled the pace and rhythm of their own labour. The factory — a centralized site assembling many workers around expensive power-driven machinery, under coordinated time-discipline and external supervision — was a genuinely new social form, emerging first in British cotton spinning in the 1770s–90s (Arkwright’s Cromford mill of 1771 is the conventional emblem) and becoming the dominant mode for industrial goods by mid-century. E. P. Thompson’s classic essay “Time, Work-Discipline and Industrial Capitalism” (1967) argued that the factory imposed a new experience of time itself — the clock, the bell, the fine, the loss of customary “Saint Monday” idleness — and that the making of an industrial working class was as much a transformation of work-discipline and consciousness as of output. This reorganization of labour is, for the rehabilitation school, revolutionary by any reasonable meaning of the word, and it is precisely the kind of structural change that a per-capita GDP figure cannot register: two economies can post the same growth rate while one is organized around the household and the other around the factory floor.

The accompanying urban transformation gives the qualitative claim a quantitative spine. England’s urban share rose from roughly 25% in 1780 to about 50% by 1850 and toward 80% by 1900 — among the fastest sustained urbanizations in human history — and the new industrial cities (Manchester’s population rose from ~25,000 in 1772 to over 300,000 by 1851) were a new kind of place, not merely larger versions of the old market towns. The point is that the form of British society was being remade in exactly the decades the aggregate growth figure looks unremarkable.

The disaggregated-data turn

The rehabilitation position has been strengthened over the past fifteen years by a body of disaggregated evidence that the aggregate series structurally omit, and the strongest of it concerns women and children. Humphries’s Childhood and Child Labour (2010), built from roughly 600 working-class autobiographies, documents how central child labour was to the industrial workforce and how the IR reorganized the timing and conditions of children’s work and apprenticeship — an experience invisible in adult-male wage series. The Humphries–Schneider spinning work (2019) reconstructs the earnings of hand-spinners, overwhelmingly women and children, and shows that this lead pre-mechanization activity was low-paid and largely missing from the standard wage indices. Crucially, the same finding cuts two ways: it supports the rehabilitation claim that aggregate measures miss the demographic majority of industrial labour, and it is the empirical basis for the Humphries–Schneider critique of Allen’s high-wage thesis — the disaggregated turn is doing double duty across the debate.

The deeper methodological alliance is with the regional-GDP and occupational-structure reconstructions (the Cambridge Group’s work on occupational structure; regional GDP series). These show that national aggregates average over divergent regional trajectories — the industrializing north pulling sharply up while the once-prosperous agrarian south stagnated or fell — so that the modest national figure is the sum of a boom and a relative decline, not a description of any actual region’s experience. Berg and Hudson’s original contention, that the mean conceals the transformation, has thus been borne out by precisely the kind of finer-grained quantitative work that the early aggregate program could not perform.

Key evidence

Sectoral growth rates — cotton at 6–7% per year; pig-iron output rising roughly ten-fold from ~250k tons in 1800 to ~2.7m by 1852 (and a hundred-fold from the ~25–30k tons of c.1720); coal output rising from ~3m tons/year in 1700 to ~15m by 1800 to ~225m by 1900. These are historically unprecedented.
Factory employment shares — by 1850, factory employment in cotton alone was on the order of 400,000 workers; total industrial-employment figures were a multiple of this. The transformation of workplace organization is quantitatively documented.
Urbanization rates — England’s urban population rose from ~25% in 1780 to ~50% in 1850 to ~80% by 1900. This is one of the largest and fastest urbanization events in human history.
Regional employment specialization indexes — systematic work by Humphries, Hudson, and others shows the concentration of specific industries in specific regions rose sharply across the IR century.
Humphries’s autobiographies database — ~600 working-class 18th–19th-century life histories provide qualitative and demographic evidence of the transformation of work and family that aggregate series cannot capture.
Berg and Hudson’s regional economic history — detailed sectoral and regional reconstructions showing the internal dynamism hidden by aggregate means.

Raw cotton imports into Britain (million lb, log scale), c. 1700–1860. Hover for year-by-year values. Sources: Mitchell, British Historical Statistics; Deane & Cole.

Data table

Year	Imports (million lb)
1700	1.99
1710	0.72
1720	1.97
1730	1.55
1740	1.65
1750	2.98
1760	3.87
1770	4.76
1780	6.77
1790	31.45
1800	56.01
1810	132.49
1820	151.67
1830	263.96
1840	592.49
1850	663.68
1860	1,390.94

Major critiques

— CRAFTS & HARLEY: you cannot wish away the aggregate data. If the IR was genuinely transformative, why did per-capita GDP growth wait until the 1830s to accelerate? The gradualist reply is that aggregate measures simply reflect the composition of the economy: an economy in which the transformative sectors are small shares must show modest aggregate growth — a description of the period, not a measurement error to be corrected. On this view the Berg–Hudson rehabilitation changes the subject (to organization, demography, regional structure) rather than refuting the quantitative finding, and the two are not actually in contradiction.

— The uniqueness claim needs comparative benchmarks: cotton growing at 6% while the rest grew at 0.3% is not self-evidently more “revolutionary” than earlier bursts of localized sectoral dynamism — Italian silk, Flemish cloth, Dutch shipping and herring, late-Ming Jiangnan textiles. The assertion that this sectoral revolution was historically singular requires explicit cross-episode comparison that the Berg–Hudson tradition more often asserts than supplies.

— The qualitative claims are hard to falsify: “the factory system was a revolution in the organization of work” is a real proposition, but it has no agreed metric — how is “revolutionariness” measured? The GDP critics have the methodological advantage of a clear, replicable number; the rehabilitation case leans on qualitative judgment that is harder to test and harder to bound.

— The welfare paradox: if the IR was the genuine transformation Berg and Hudson claim, why did working-class real wages and adult stature fall through ~1780–1830 (Feinstein’s wage series, Komlos’s anthropometric work)? Their answer — that the structural changes set up the post-1850 welfare gains — is a downstream-benefits argument, not an answer to the short-run welfare critique, and it concedes that on the measurable contemporary metrics the “revolution” was, for most people who lived through it, immiserating.

— From the gradualists: at bottom Berg and Hudson are restating an older “revolution” rhetoric in newer social-historical vocabulary. Once the aggregate numbers are conceded — and they are — whether to call the residual structural changes a “revolution” is largely a matter of emphasis rather than a finding that can be adjudicated.

Status

Contested, and substantively important. The Berg–Hudson rehabilitation has been influential among social and regional economic historians and has shaped subsequent work on women’s labour (Humphries), material culture (Berg), regional industrialization (Hudson), and the Anglophone “history of capitalism” school (Beckert and others). Among quantitative economic historians centered on GDP and productivity measurement, the position is less widely embraced — the Crafts tradition remains the default framework. The empirical ground shifts under both positions: newer disaggregated data (Humphries–Schneider on spinning, Kelly-Mokyr-Ó Gráda on skill content, regional GDP reconstructions) generally support the Berg–Hudson reading that aggregate measures systematically miss the transformations that matter. The strong form — that Crafts–Harley is simply wrong — has fewer defenders than the weak form, that both quantitative and qualitative analyses are needed and that the qualitative has been under-represented in recent economic history.