REWT Rivers of England and Wales, Temporally

Aims

What this produces

A routable river network for England and Wales, published as a series of dated cross-sections, with modelled flow on each and the evidence for every line carried with it.

Routable means what it says: from any stretch you can follow the water downstream and reach tidal water. That is not true of the modern published network, and making it true is most of the work.

The name is a pun on route, and it is meant seriously. A river network that cannot be traversed cannot answer the questions rivers were used for, and those questions are largely economic: what could be moved, from where, and at what cost. River improvement in the late seventeenth century was undertaken above all to get coal to towns where hauling it overland made it ruinously dear. So where a river ran and when it could be worked is an input to the history of commerce and industrialisation, and not only to the history of landscape. Navigability is a property of a route; this is the only open historical river dataset being built to carry one.

And a route is only half of what a river was for. Every reach in the network carries a modelled flow and a fall, and flow times fall is power. Before steam that is what a river did for the place beside it: grinding corn, fulling cloth, driving hammers, bellows, saws and eventually machinery. Power potential is as computable from this dataset as navigability, and because both sit on the same lines a question becomes askable that neither answers alone — did a place grow because of what it could ship, or because of what it could make? Before coal made the first decisive, the second may have mattered more. What that model is and is not is set out with the method: it is a screening surface rather than a site assessment, and its most useful output is where it disagrees with the mills that were actually built.

Who it is for

On that last point, because it is the reason the first edition is worth publishing at all. The open national product, OS Open Rivers, ships a topologically structured link-and-node network and is the base this project builds on — but Ordnance Survey’s own documentation says detailed analysis is not supported in it and directs anyone wanting that to a licensed product. This project’s measurements agree: the topology is there and the traversability is not. The open datasets that are routable at national scale are global and derived from elevation — HydroRIVERS is extracted from a 15-arc-second gridded surface — which is a hydrological model of where water would run, not a record of the channels that are there.

The closed products are the other way round. The Environment Agency’s Detailed River Network, its successor the OS MasterMap Water Network Layer, and the UKCEH Digital River Network are all derived from survey, not from a terrain model: the EA network was captured from OS MasterMap and built into a topologically correct network with automated rules and field survey, and the UKCEH network was digitised from OS 1:50,000 mapping and then used to define the flow paths in CEH’s terrain model rather than being derived from it. So the gap this project fills is not methodological. Good routable survey-based hydrographies of England and Wales exist. What their licences restrict is re-use, derivation and publication — which is the whole of what this project does.

And every one of them is a map of the present. So is the first edition here: a 2026 snapshot, the modern survey repaired, offering nothing a licensed product does not except that anyone may build on it. What is new begins with the second — the same network at other dates, reconstructed, each edition a cross-section rather than an update. A snapshot is a thing you can already buy. A dated series is not.

Two axes of publication

These are independent, and confusing them would mean publishing nothing until the research is finished — which is to say never.

Editions — how complete the evidence is

   
First the modern survey repaired into a working drainage network. Makes no historical claim, and ships as soon as it is sound.
Second old courses recovered from map evidence — channels labelled as superseded, and valleys drowned by reservoirs. Bounded, countable work.
Third and after courses recovered from documentary and cartographic research, released incrementally, a course at a time. This part never completes.

Epochs — which date is drawn

Each epoch is a separate build, not a filter: flow is modelled on the network, so changing which channels exist changes where the water goes. Seven are proposed, each chosen to sit between phases of change rather than mid-transformation:

1086 · 1300 · 1540 · 1600 · 1700 · 1830 · 1900

1600 is the one that matters most — the last state before the Fens and Humberhead were rebuilt. The first edition draws none of them: it is the repaired modern survey, dated as what it is, and the historic cross-sections are worked back from it.

An epoch is published when its evidence supports it and not before. An empty medieval map is worse than no medieval map, because it looks like a finding.

Publication

A paper will announce the first edition, which settles its timing: after the modern network is sound, and not waiting on the documentary research behind the later ones. Venue and collaborators are not yet decided, and this page will name them when they are rather than before.

What the paper is about is decided, and it follows from the navigation strand: a traced, dated, routable watercourse network is an input to the economic history of movement — what could be carried, from where, at what cost — and that is a literature with its own established datasets and its own people. The natural collaborators are the ones already working on transport and industrialisation, and the contribution offered to them is the channel rather than the question.

And there is a specific invitation in that. The most substantial body of work on English and Welsh inland navigation — dated section by section across three and a half centuries, with the pre-drainage Fen channels traced on their own courses — is not openly published, and its authors are free to publish it as they wish. This project is built to be the vehicle for exactly that: what each side has, and why this is a vehicle rather than a competitor, set out plainly. The offer is an exchange — topology, direction, flow, an evidence-first date model, and an open licence enforced by the exporter rather than promised — and it is an offer rather than an assumption.

And it is meant to bring people in. The paper is as much a solicitation as an announcement. The scale of the work sets out which strands can be finished in house and which cannot, and the ones that cannot — the mill channels, and the regional documentary research — need people who already know a region or a record class and will be faster at it by an order of magnitude than anyone learning it. Contribution is wanted at both ends of that range: a reader who can correct one stretch of river matters as much as a collaborator who can take on a county, and the tooling is built so that either arrives attached to a specific line rather than as a remark about a map.

What this is not