skip to primary navigation skip to content

Transport, urbanization and economic development in England and Wales c.1670-1911

Transport, urbanization and economic development in England and Wales c.1670-1911

Reconstructing the flows of goods around the economy of England and Wales 1670-1911

One key plank of our Leverhulme/NSF project is to model the flows of goods and services around the various transport networks at three key dates: c.1670 before the turn-piking of the main roads, the extension of navigable rivers and the building of canals; c.1830 when the foregoing was essentially complete but before the railways and iron and steel-hulled steam-powered ships had become significant; c.1911 when the railway network and coastal shipping were close to their peaks. Since we wrote our grant applications in 2012 we have become aware of a vastly larger body of data published among the millions of pages of parliamentary papers on the flows of goods around the economy than we were aware of then. These data would allow us to model the flows of goods along every main road, along rivers, canals, the coast and railway lines far more fully than we realised was possible in 2012. We will need further funding to exploit some of this material and to explore further possibilities more fully. In addition to the body of data contained in the parliamentary papers we have also become aware of the astonishing detail and volume of material available in manuscript sources, especially in relation to coastal shipping. Some of these sources underlie the published data but provide much detail lost in the published aggregation – right down to the cargoes of individual ships moving goods between specified ports in some cases. Other data were never published in any form, including much of the eighteenth century data on coastal flows and data relating to loading, unloading and journey times between specific pairs of ports. We have applied to the Isaac Newton Trust for funding for a pilot project to enable us to apply for larger scale external funding for a much more comprehensive exercise.

A more detailed account of what we like to do follows. The aim of this document is to provide an indication of some the new sources we wish to use and how we would plan to use them. Since this document is being written in advance of any comprehensive search of even the parliamentary papers, let alone the manuscript material or other printed sources such as newspapers, what is mooted here represents only a starting point and it is likely that we can do considerably more than is suggested here. The document is structured as follows. Section one provides some background. Section two discusses the dominance of coal in freight traffic. Section three discusses what could be done using printed sources with a primary focus on what is possible in 1911. Section four briefly discusses some of the manuscript material available. Section five gives an indication of what we would hope to achieve with the pilot funding we have applied for.


On our current Leverhulme and NSF funded project, Transport, Urbanization and Economic Development 1670-1911 we have built on previous projects to create a multi-modal transport GIS which integrates the main roads, turnpike roads, navigable rivers, canals, railways and coastal shipping routes. Our transport GIS is dynamic, that is to say we can map the networks for any date between 1670 and 1911. This allows us to model journeys around the transport system using any combination of networks in this date range with journeys optimised on speed, cost or (for passenger journeys) comfort. The Leverhulme project includes a modest sub-project aimed at a partial reconstruction of the flows of people and goods around the network at four benchmark dates: 1670 (before the improvements to rivers, the development of turnpike roads or the building of canals) 1830 (at the peak of the canal, river and turnpike networks but before the railways and steam shipping became important, 1870 (for reasons of data availability) and 1911 (at the peak of the railway network but before the long-term decline of coastal shipping consequent upon the first world war)[i].

Since we submitted the Leverhulme and NSF funding applications in 2012 we have become aware of two vast bodies of source material which could be mined for data which would allow us to attach quantified flows of goods, passengers and (for some parts of the network including railways) freight and passenger revenues to each segment of the transport network at various dates. One set of sources consists of the millions of pages of parliamentary papers produced in the period, now all available as PDFs online through Chadwick-Healey's Parliamentary Papers Online. The other is a dauntingly voluminous body of manuscript material at the National Archives and other archives around the UK and beyond. Much of the latter contains disaggregated data that was aggregated for publication in Parliamentary Papers but it also includes much material that was never published. At the time we submitted our applications in 2012 we were aware of only a fraction of the available materials in parliamentary papers and of almost none of the material available in manuscript form. Moreover, new digitised datasets, not available in 2012, relating to occupational structure have become available for every census parish in 1851, 1861, 1891, 1901 and 1911. As a result of these developments we now have the opportunity to model the flow of goods and people around the economy in a much more comprehensive way than we previously realised was possible and would like to make this the central focus of a new follow-on research project. Moreover, we want to focus on the flows of coal around the economy for four reasons: (i) the transformation of the energy regime entailed in the massive increase in coal use over time and its rise to dominance in energy supply underwrote the Industrial Revolution and the associated changes in economic geography but was substantially dependent on developments in transportation systems (ii) the need to move large quantities of coal around was a key driver of transport improvements in the period (iii) coal dominated the ton mileage of freight goods around the economy throughout the period (iv) the movement of coal is much better documented than any other commodity and is relatively easy to reconstruct. To achieve our aims here we need some interim funding so that we can (i) explore and assess the voluminous printed and manuscript source material available (ii) experiment with building a 'first-pass' model of the flow of goods around each section of each transport network (iii) evaluate both what is possible and where the best returns to effort are the be had so that (iv) we are in a position to write a properly costed and compelling large research grant application to the ESRC for a much more comprehensive project.


We begin with the dominance of coal in ton-mileage. Dorian Gerhold calculates that in 1826 the ton-mileage of goods coming into London by road was around 19.25 million ton miles. John Armstrong estimates that in the same year the ton-mileage of coal coming into the London by sea from the north-eastern coal field alone was around 600 million ton miles (two million tons coming three hundred miles). Shipments of East Anglian grain accounted for perhaps another 20 million ton miles[ii]. This illustrates two key points. First, at this date it is likely that the total ton mileage of freight in the economy was dominated by coal. This would also have been true at our terminal date of 1911 and was probably true in 1670 too though some further work would need to be done to confirm that. From this it follows that if we can document the flow of coal around the economy we will be able to document well over half of all the freight traffic measured in ton miles. This is fortuitous because the movement of coal is particularly well documented. The second key point is the central importance of coastal shipping which, as Armstrong noted, was the dominant freight transport mode over longer distances across the eighteenth century and much of the nineteenth. Nor was shipping rapidly pushed aside by the rise of the railways. Even as late as 1911 coastal shipping still brought in around half of London's (vastly increased) annual supply of coal. Since this coal came predominantly from the north-east while the other half, supplied by rail, came from the Midlands and Yorkshire, coastal shipping still accounted for over half the ton mileage in the largest single component of freight flows, supplying London with coal.


Turning now to the modelling of flows in 1911, we can begin with what we planned to do on our Leverhulme grant. We alluded in the application to the possibility of using the Railway Returns to model the flow of good and passengers around the rail network in 1911 but we were uncertain how much could be achieved and explicitly noted that this was one element of the project that we might have to cut back in order the focus on more central components of the Leverhulme project. The Railway Returns, which were published annually from 1870 and more intermittently before that, contain tabulations of aggregated data for each railway company. These included: (i) the number of miles of line (distinguishing double and single tracked lines; (ii) the tonnage of minerals with the tonnage for coal listed separately from other minerals; (iii) the tonnage of other merchandise transported (but not the ton mileage or over which miles of that company's network the goods were moved); (iv) the number of locomotives; (v) the number of carriages; (vi) the revenues from the mineral and general merchandise and also from first, second and third class passengers and from season ticket holders; (vii) the tonnage of coal used and much else. The first step to linking this flow data to our rail GIS will be to attach attribute data with the company name to each section of our rail GIS. Only then can the data we would digitise from the Railway Returns be linked to the GIS. Some of this we can accomplish with our present funding and this is likely to be sufficient for us to model the flow of passengers around the rail network in 1911 (with certain limitations which we could move beyond if the pilot is funded). We will be able to compare this with the volume of passengers by stage coach in 1830 and c.1680, all within the Leverhulme project.

However, our capacity within the current project, to model the freight flows on the railways is much more limited, not least because we don't know which sections of the company's network the known tonnages travelled along, and over what distances they travelled, while our capacity to model the vital flow of goods along the coast between the various ports is non-existent. Moreover, we currently have no data on the flows of goods on any non-rail networks after c.1830. The new sources we now propose to use would make it possible to model the flow of goods on railways much more fully and accurately and also to model the flows of goods on other parts of the transport network over the period c.1830-1911 and probably earlier.

Before describing the main sources to be used, it is important to note that these sources each give a partial but detailed and quantitative picture. All the sources taken together still won't provide direct data on the flows of goods on every section of each network but they will allow for very constrained estimates where direct data are not available. In the case of coal, data are available for the tonnage produced on each coal-field each year and for other minerals the sources document the tonnage for each and every mine. The available data would allow us to: (i) estimate parish level production figures for coal to sit alongside the hard mine-level data on other minerals (ii) partly measure and partly estimate quantities of these minerals consumed in each parish (or the ports from which they left the country); (iii) measure the flows of coal, minerals and other merchandise along some sections of the transport network (iv) estimate the flows of goods along the remaining segments of the various transport networks. Each additional source we add to the mix, further constrains the estimates of where coal and other goods flowed and it is our anticipation that where estimates are necessary these will generally be constrained within tight limits. For instance, where we have data on: coal production on a coal field; flows of coal out through nearby ports and flows of coal out of the coalfield by railway and by canal, we can make a very constrained estimate of the coal consumed locally. Since we: can: in the same way calculate the retained iron ore in the vicinity; document the locations of blast furnaces within the vicinity and the size of the associated labour force; have aggregate data available for national fuel consumption by blast furnaces and have county level data on the horsepower of steam engines in blast furnaces we can make parish level estimates of coal consumption in blast furnaces. Similarly we have evidence of the installed horsepower in factories and mines (at county level and sometimes by town) but have parish level occupational data. We can therefore make estimates of coal consumption by parish in other industries. We also know the population of every parish and can make estimates of domestic coal consumption. Moreover, in 1911 some 8% of coal, and in the 1850s and 1870s some 3% was consumed by the railways themselves. By default, information on the movement of coal, freight and passengers provides information on the geography of this usage. Since for some dates we have contemporary statistical estimates by the state of consumption in different branches of industry and for domestic use, these estimates are constrained further.

The available data would allow us to estimate with a high degree of accuracy parish level production figures for coal to sit alongside the hard mine-level data on other minerals. The data also identify UK coal exports, which determine the tonnage of coal available for domestic consumption. Our data on parish populations occupational structure and other data (such as the locations of blast furnaces) can be used to estimate the amount coal consumption across parishes. Warde's work on coal consumption by industry in 1907 will help further with districting coal consumption. The second step is to estimate the flows of coal between points First, we can use our multi-modal network to identify the lowest transport cost between every parish and the set of producing parishes. The producing parish with the lowest transport cost would be a reasonable first-pass candidate as the origin of the consuming parish's coal, and by implication we would have a candidate route by which the parish received its coal. In a more sophisticated model we would identify the overall set of flows that would have minimised overall transport costs Proceeding with this method we can construct a national network of flows by rail, by coast, by inland waterway, and by road. Second we would check how our candidate flows compare with observed flows of coal by rail, coast, and inland waterway network. Note the observed flows are partial but extensive and will be valuable in making corrections to flows proposed by our network model. Through an iterative process between hypothesized flows and partial data on observed flows we can measure with a good deal of accuracy the actual flows of coal on the network.

We can now turn to the key series of parliamentary papers to be used. First of all, the Mineral Statistics, which were published annually from 1853, provide each year data on the tonnage of coal produced in every coal field in Britain and a full list of all collieries on each coal-field. For all other minerals, output data are available annually on a mine by mine basis (together with the value of that output). A full list of iron and steel blast furnaces (distinguishing those in blast and not in blast in the year in question) is also available annually. This means that for all minerals other than coal, we could, once we geo-reference the data, identify the tonnage of output by parish each year directly. For coal, whilst we mostly only have direct evidence of output for whole coal-fields, we do have data from the new census datasets of the numbers of miners and others employed in coal mining in each parish in each censal year, meaning that we can distribute coal-field tonnage to parishes in relation to the labour force. Whilst this is imperfect, it unlikely to produce any very large errors and is a very considerable improvement on what of output data in proportion to the parish labour force works in practice and, perhaps, to improve it. The Mineral Statistics also provides data annually on the tonnage of coal taken by particular rail companies from might otherwise have been possible. Moreover, for some years, there are output figures available on a mine by mine basis on some coal fields[iii]. This will allow us to see how well our model for the districting each coal field and partial data on where it was delivered to. In some years the tonnage of coal from individual mines in various coal fields which made its way to London is documented. From 1819 to 1828 there are published data (and these data may be available for parts of the eighteenth century too) on the tonnage of coal annually entering and leaving every customs port in Britain and distinguishing between the coastal, Irish and foreign trade[iv]. Before the late nineteenth century this would have accounted for the great bulk of all ton mileage in the economy. From 1853 the same data were published annually in the Mineral Statistics. They may well be reported elsewhere in the intervening years. While there is no port-to-port data, this can be modelled on a number of assumptions consistent with the data[v]. Exporting ports were generally at some distance from the coal mines served and every ton of coal flowing out of that port can be assumed to have travelled from the nearby coal-field by road, rail, canal or river. In addition to being able to estimate the parish level production of coal and all other minerals, we can use the occupational and population data in the censuses to estimate the consumption of coal in every parish. This process will be further facilitated by the data on installed steam power in factories and workshops in the Factory Returns published annually from 1868 and intermittently and with less data from the 1830s (though factories are not generally located more precisely than the county) and the parish level lists of blast furnaces in the Mineral Statistics. In previous work Warde has undertaken a complete reconstruction of fuel consumption in British industry and utilities (disaggregated into some 150 branches) from the Census of Production of 1907, which can be related to both output and workforce. In the case of highly spatially-concentrated industries this information provides an additional constraint on likely demand. More generally, we can link industrial consumption per worker to occupational data to provide a rough indication of localised industrial demand as a further constraint and test the plausibility of this method of assessing the distribution of coal consumption in periods where occupational date provides by far the richest and most fine-grained source available but archival information also is available on the rate of fuel consumption to worker in specific firms. Much further data are available in the parliamentary papers. For instance, the 1871 Royal Commission on coal addressed coal use for intense heat in a range of metallurgical industries, as well as potteries, brickmaking and glassmaking, and provided estimates for the distribution of all coal use across the sectors of industry and in households[vi]. Data concerning tonnages carried on many canals and navigations for 1828, 1838, 1848, 1858, 1868 were published in 1870[vii]. , The Canal Returns of 1888 and 1898 give tonnages of local and through traffic. The eleven volume Royal Commission on Canals and Waterways provides detailed information concerning the carrying capacity of vessels using particular canals and rivers section by section. It also supplies mileage, capital, tonnage, gross and net revenues for major routes in 1905. Further data are available in other printed sources. For instance from 1757 onwards Lloyd's Evening Post (and it successors) reported the names of ships leaving and entering British ports for and from foreign parts and for London in the coasting trade, on a daily basis (though we have not yet checked the completeness of these data).

It will be apparent from the foregoing that we have yet to work out exactly how to do all this in detail and that that there is far more material available than we could realistically hope to utilise. This is precisely why we are applying for interim funding now. Before we apply for a further tranche of major external research funding we need to (i) digitise some of the available data; (ii) search more fully the printed parliamentary papers and other printed sources (such as trade newspapers of which The Colliery Guardian looks particularly promising) which will no doubt throw up further sources of data which we are not yet aware of; (iii) experiment with ways of putting all the data together to estimate the flows of coal; other minerals and general merchandise round the network which will provide proof of concept but will also serve to identify key evidential lacunae (iv) explore some of the available manuscript sources (see section III) which will help to (v) test the model and (vi) identify ways of filling important lacunae (vii) provide other useful data. . We would probably start with the flow of coal and other minerals in 1911 when the data are most complete. In a second stage, resources permitting, we would attempt to model the flow of other goods around the transport networks. The Railway Returns, the Factory Returns and the Canal Returns and the parish level census data would again play a key role here. In addition we would use the data available in the Annual Returns of Navigation and Shipping across the second half of the nineteenth century. From 1873 these contains data on the number of ships and the aggregate tonnage entering and leaving every customs port in the United Kingdom each year, distinguishing between sailing vessels and steam vessels and between the coastal and overseas trade as well as between ships in cargo and ballast. Somewhat different data are available annually prior to 1873. It should be noted that these 'tonnages' are volumetric capacity measures not the tonnage of the cargoes. This would have to be estimated and much work will be required to develop plausible and robust methods for doing so. Having done so, it would, in principle, be possible to subtract the known totals of coal in each of these flows to arrive at estimates of the tonnage of other cargoes. Again some cross checking will be possible against sources with more detail available.

In some years further data on freight flows are available which would allow further refinements of the model. For instance, an 1828 Parliamentary paper Grain, Malt and Flour documents, for each year 1824-1827: the volume (in quarters) of barley, malt, oats, peas and beans, rye, wheat, barley meal, oat meal, wheat meal and flour, into and out of every customs port in England Wales and Scotland via the coasting trade; the same flows into and out of every British port to and from Ireland (and the flows into and out of each Irish port), and the imports/exports to 'foreign parts.' Again, there is no port-to-port data, but this can be modelled on a number of assumptions consistent with the data. Thus for the year 1827, for instance, we can estimate the two largest commodity flows into and out of every customs port in England and Wales. Similar data are available for 1844[viii]. Quite possibly further data are available for other years.

Each 'custom port' in reality represents a number of distinct locations along a given stretch of coast line so in practice this a first approximation. However, there are sources for certain dates, which would allow us to assess the relative importance of the various 'shipping places' or creeks, harbours and ports covered by each customs port. These include the 1882 Return of Harbour Authorities, which documents all 634 'shipping places' in the United Kingdom in 1881 and the Customs Ports to which they belonged and the Pilotage Returns for 1910-12 gives the number of British and foreign vessels in both the coasting and overseas trade for every port and harbour which employed a pilot over the three years. Once each port and harbour is assigned to their appropriate customs port tonnages from customs ports can be interpolated across the ports and harbours using the numbers of vessels as a guide to their relative importance. As with so many aspect of our work on freight flows, what is documented here relates to what we know about now, but one outcome of the pilot project would be an increase in our awareness of sources.


One further element of the pilot project would be to assess some of the voluminous archival material in The National Archives, The Maritime Archives and Library (Liverpool), County Record Offices, the British Library and elsewhere. This material would be scoped during the project pilot study both to fill key lacunae for the 1830-1911 period and with an eye to creating datasets for the 1700-1830 period that mirror published data available in the later period. John Hatchers' The History of the British Coal Industry. Before 1700. Volume I (1993) and J.U Nef's, The Rise of the British Coal Industry (1932) both contain considerable data in relation to shipments of coal in the sixteenth and seventeenth centuries. The available material in the eighteenth and early nineteenth centuries is much more voluminous and we have only just begun to explore the possibilities here.

Two manuscripts in the British Library which summarise coastal and foreign shipping tonnages in and out of British ports between 1709 and 1779 are likely to be key sources. These documents are abstracts from Customs House records destroyed by fire in 1814.[ix] Further available material includes very detailed data on individual shipments, such as the Maritime Bills of Entry used by Valerie Burton to document the trade of Liverpool in the 1850s[x]. The data record, ship by ship, details of the cargoes being unloaded at individual ports right down to the number of boxes of eggs, firkins of butter, numbers of livestock and so on. The available volumes include: Liverpool (1820-1939), London (1779-1939), Bristol (1832-1917), Hull (1831-1939), Dublin (1850- 1923), the Tyne (1875-1927). These kind of sources, collected selectively, would provide detailed data that would allow cross-checking of some of estimates deriving from published sources, discussed above, against actual data, which would allow iterative improvements in methodology. Similar data are available on the seventeenth and eighteenth centuries in the Port Books Coastal (E190) and the Coastal Trade Books (E122) at the National Archives. Both sources provide considerable data on the goods shipped for individual port to port shipments 1650-1789, and can demonstrate coal shipping in relation to myriad other goods transported on the coastal network, such as agricultural produce, consumables including tobacco and spirits, and timber. The availability of comprehensive data for the range of goods transported around Great Britain over this long period allows for unprecedented study of coastal traffic in general, while demonstrating the relative importance of coal transportation overall.

Another example of manuscript material are the 'Crews Lists' available from the 1830s onwards and contained in 26,000 boxes of documents (10% in the TNA, some percentage in County and Local Record Offices but approaching 90% now held at the Memorial University of Newfoundland)[xi] which contain details of individual voyages of ships (and all crew members) allowing us to calculate journey times between pairs of ports as well as times spent in port loading, unloading and waiting for the right weather. It is most unlikely we would to do more than sample a subset of the material remaining in the National Archives. Similar data on journey times can be derived from the Coast Trade Books contained in E122 at the National Archives for the period 1650-1789. Similar data on journey times can be derived from the Coast Trade Books contained in E122 and from the Port Books Coastal at the National Archives for the period 1650-1789. These sources reveal completed coastal sailing voyages from 1650 that go unrecorded elsewhere, partly as coastal shipping was unscheduled, departures being contingent solely on weather and other events on the ground.

A further set of sources, not yet investigated are the Records of the Coal Factors' Society at the London Metropolitan Archives. These begin in 1761 and consist of financial records including ledgers and cash books, factors' day and entry books, factors' and meters' postage books, an alphabetical list of ships giving gross and net tonnage, lists of named ships with total monthly cost to each factor, comparative lists of the highest prices of best house coals taken for each.


Our long-term goal is to create parish level origin and destination flows for coal, other minerals, general merchandise at people at a number of dates with a key focus on c.1670, c.1830 and c.1911. We may wish to add further benchmarks at some stage with one around 1770 (by which time there was a national network of turnpikes but not yet a truly national canal network) being an obvious desiderata. It might also prove necessary to creating a benchmark in the early 1850s as a stepping stone to creating an 1830s benchmark. Clearly what we can do in a short-term pilot project is limited. Our intention would be: (i) undertake a substantial scoping exercise of both published and manuscript c1670-1911 (ii) to digitise a spatially comprehensive body of flow data c.1911 and some very selective data (for other benchmark years (ii) to try to build a first-pass comprehensive model of transport flows c.1911 focussed on what will be most helpful in making the case for the planned follow-on application) (iii) model some carefully selected sub-sets of the transport flows at earlier dates (iv) to test estimated flows against more spatially disaggregated data where these are available on specific sections of the networks at one or more dates (v) write a number of working papers and/or journal articles deriving from this work (vi) use all of the foregoing as the basis for an application to the ESRC for a much larger research project on transport, energy and the transition to coal c.1670-1911.

[i] Armstrong, J., 'Climax and climacteric, The British coastal trade 1870-1930', in Starkey, D.J., and Jamieson, A.G., Exploiting the sea: Aspects of British maritime history since 1870 (1998)

[ii] Armstrong, J, The Vital Spark, pp.17-18

[iii] For instance, the 1830 House of Lords Select Committee Report on the State of the Coal Trade. Contains mine level data for 35 colleries on the Tyne and 8 on the River Wear in 1828. The former accounbts for over 90% of the coal shipped out on the River Tyne in that year. For four of these colleries data are provided for a run of years giving a handlke on annual variation per mine. See pages 8, 29 and 30. Mine by mine output data for the Tyne and the Wear colleries is also available for 1842/3 in the 1843 Report of Royal Commission on the Midland Coal Field along with data on the installed steam-power capacity for both drawing and pumping; pit-head prices and employment data, pp. cvii-cviii.

[iv] Data for the years 1819-1828 were published in the 1830 House of Lords Select Committee Report on the State of the Coal Trade.

[v] Occasionally data were published on port to port shipments and such documents would allow us to test our model. See for instance the 1808 paper, Account of Coals imported into the ports of Bridgwater, Bristol and Gloucester. One component of the pilot would be to search more systematically for such material in the parliamentary papers as well as scoping the manuscript material from which such datasets could, with sufficient labour inputs, be reconstructed.

[vi] Report of the commissioners appointed to inquire into the several matters relating to coal in the United Kingdom, 1871, Appendix E.

[vii] BPP Navigation and canal companies. Returns relating to inland navigation and canal companies in England and Wales (1870).

[viii] 1845, Return of Shipping … laden with foreign grain … and clearing coastwise.

[ix] British Library: Add. MS. 11256 and Add. MS. 11255: These were compiled by J. Dalley, assistant registrar of shipping in 1803For an example of the usage of these data for south-west England, see: Fisher, S., 'Devon's maritime Trade and shipping, 1680-1780' in Duffy, M., Fisher, S., Greenhill, B., Starkey, D.J.,Youings, J., (eds.) The New Martime History of Devon.

[x] Burton, V., 'Liverpool's mid-nineteenth century coasting trade in Burton, V., (ed). Liverpool shipping, trade & industry (1989) pp. 26-68

[xi] Joanna Thomas, unpublished conference paper, 'A mountain of documents: Bias in archives and sampling.' Bias in History Conference, 15th May 2016.