01Executive summary
Four milestones in which the compact high-pressure engine and the iron hull between them end the age in which transport meant wind, water and muscle.
John Wilkinson launched the first iron vessel in 1787 against the settled conviction that an iron ship must sink. Trevithick put a high-pressure engine on rails in 1804. Robert Fulton's Hudson River service of 1807 is the point from which steam navigation dates as a commercial reality. And the Rainhill trials of 1829 settled the form of the locomotive around a multitubular boiler and direct drive.
02The iron hull: an argument that had to be won twice
Iron is about eight times as dense as water, and the intuition that an iron ship must sink is extremely strong. It is also wrong, and had been demonstrably wrong since Archimedes: a hull floats if the weight of water displaced by its enclosed volume exceeds the weight of the vessel, and the density of the material is irrelevant except through how much of it is needed.
Wilkinson's barge of 1787, roughly 21 metres long and built of riveted iron plate, is therefore not a triumph over physics but over a widely held misconception. What made iron construction genuinely attractive was not that it floated but what followed from it.
Thinner structure, more capacity
Iron plate provides the same strength as timber at far less thickness, so an iron hull of given external dimensions carries a substantially larger cargo.
Length is no longer limited by trees
Timber ships are limited by the length of available timbers and by hogging of long wooden hulls. Iron plate and frames can be assembled to any length.
It does not rot or burn
Two of the main causes of loss and of maintenance cost in a wooden ship are simply absent.
New failure modes arrive with it
Corrosion, fouling, and compass deviation from a large mass of iron are all new problems created by the material, and each needed its own solution.
The last card is the recurring point of this whole set: substituting a material removes the problems of the incumbent and introduces a set of its own, and the new set is not visible until the substitution is made.
03Steam on rails: why 1804 and not earlier
Rails are older than steam. Wagonways carrying horse-drawn trucks on wooden and then iron rails had served collieries for a long time, for the same reason a canal works: reducing rolling resistance lets one animal move a great deal more. What was missing was a motor that could travel with the load.
A Watt engine cannot do this. It has a condenser, a supply of cooling water, a heavy beam and a building around it — and it develops modest power for its enormous mass. Trevithick's non-condensing, beamless, direct-acting engine has none of that, and putting it on wheels in 1804 was the immediate application of the previous part of this series. The locomotive is a consequence of high pressure, not a separate invention.
A stationary engine can weigh whatever it likes; the foundation does not object. A vehicle must carry its own engine, so every kilogram of machinery is a kilogram not carried as payload and a kilogram that must itself be accelerated and hauled up gradients. Transport applications are decided on power-to-weight, and this is why the condenser had to go before rail traction was possible. The same reasoning appears in the aeroplane, in the gas turbine and in the electric vehicle: the question is never how much power, it is how much power per kilogram.
04Fulton: the first commercial application
Robert Fulton's Hudson River operation of 1807 is where the steamship dates from as a commercial success. It is expressly not where steam navigation was first demonstrated. James Rumsey of Virginia tried jet propulsion during the 1780s and put a vessel on the Potomac that made about 6 kilometres an hour against the current. Others experimented with considerable success, and with decided influence on Fulton.
This series and the three that follow are consistent on this point, and it is worth stating once more here. Rumsey and others made steam vessels move. Fulton made steam vessels a business, which requires reliability, a route with paying traffic, capital, and an operation that can be repeated every day. Those are different achievements requiring different abilities, and conflating them produces both bad history and bad engineering judgement. The same distinction separates Kilby from Noyce, the first ARPANET packet from a working network, and a landed booster from a reflown one.
Steam at sea also arrived in stages rather than at once. Early steam vessels carried sail as well, because engines were unreliable, coal consumption was heavy, and a vessel that could not complete a passage on its bunkers had to be able to complete it another way. The full displacement of sail took most of the century that follows, and it is covered in the next series in this set.
05Rainhill, 1829: settling the configuration
By the late 1820s locomotives worked, and it was not settled what one should look like. The Rainhill trials were competitive trials of locomotive engines on rails, and their function was to resolve that by demonstration rather than by argument. Several competitors ran, including Burstall and Braithwaite and Ericsson — the latter later known for the steamship Great Eastern.
The configuration that won combined two things. The multitubular boiler runs the hot furnace gases through many small tubes surrounded by water rather than through a single large flue, which multiplies the heating surface enormously and therefore the rate at which steam can be raised. And the cylinders drove the wheels directly, with no beam and no gearing.
| Arrangement | Heating surface | Consequence |
|---|---|---|
| Single large flue | The area of one large tube | Steam raised slowly, so sustained power is limited by boiler capacity rather than by the engine |
| Many small tubes | The summed area of many tubes — far greater in the same volume | Steam raised rapidly, so the engine can work continuously at high output in a small, light boiler |
Increasing surface area within a fixed volume by subdividing the flow path is one of the most transferable ideas in engineering. It is the same principle as a finned heat exchanger, a shell and tube condenser, a membrane module and a catalytic converter, and it is worth recognising as a general move rather than as a locomotive detail.
Rainhill is worth attention as a procurement mechanism, not just as a race. Publishing a specification, inviting entrants to build to it, and settling the question by measured performance under agreed conditions is a way of resolving a technical dispute that argument cannot settle. It also produces comparable data on several designs at once. Type testing, competitive prototype evaluation and benchmark trials in every field since work on the same principle, and Rainhill is one of its clearest early instances.
Peter Cooper used gun barrels for the fire tubes of his small locomotive Tom Thumb in 1830, which is a good indication of how quickly the multitubular idea spread and how improvised its early execution was. Within a decade the configuration settled at Rainhill was the standard, and the first of the three series that follow this one opens in a world where it is simply how a railway works.
06Takeaways for current practice
- Transport is decided on power-to-weight. The condenser had to go before the engine could travel with its load.
- Subdivide the flow path to multiply surface area. The multitubular boiler and every heat exchanger since.
- First to work and first to pay are different achievements. Both deserve credit and they should not be conflated.
- Substituting a material brings a new set of failure modes. Iron hulls do not rot and do corrode, foul and deflect a compass.
- Settle a contested question by measured trial. Rainhill produced comparable data on several designs and ended an argument that could not be won on paper.
This series stops around 1840, with the locomotive settled, iron structural, high pressure accepted and materials being tested before use. The next series picks up in 1845 with the Bessemer converter, the American system of manufactures and the beginnings of urban sanitary engineering — and with electrical engineering, which does not exist anywhere in the five thousand years covered here.