What were the alternatives to using a Boulton & Watt rotative engine? The first recorded application of steam in the Lancashire cotton industry was a Newcomen-type atmospheric engine and waterwheel at Shudehill Mill in 1783. The idea of this arrangement was that water would flow from an upper reservoir over a waterwheel, which conveyed motion to the machinery inside the mill, before entering a lower reservoir from which the engine pumped (or recirculated) it back to the upper reservoir. The advantage of this system was that it was relatively simple and reliable, removing the risk of drought bringing the waterwheel to a halt.

Science Museum, London/ Science & Society Picture Library

A model of Watt's 'Old Bess' engine, erected at Soho, Birmingham, in 1777. The engine recirculated water over the waterwheel (left), and initially used steam expansively. This meant cutting off the steam supply to the cylinder early in the working stroke and relying on the expansion of the steam thus trapped to drive the piston. This at first created a rather violent motion, resulting in the engine being named 'Beelzebub'.

These hybrid arrangements were supplemented elsewhere by traditional waterwheels, which benefited from a wide range of improvements. Inefficient undershot wheels were replaced by overshot or breastshot wheels, which made use of gravity instead of the impact of water on the wheel. The engineers John Smeaton and John Rennie pioneered the use of iron instead of wood in wheel construction, while Thomas Hewes developed the suspension wheel. Driving the machinery via a pinion drive on its rim (as opposed to the main axle of the wheel itself), this allowed waterwheels to be more lightly built, paving the way for an increase in their size after 1800.

Water was largely cheaper than steam power. There were no fuel costs (although dues had to be paid for the privilege of using a stretch of river), but the greatest savings were in the initial construction. As late as 1850, Watt-type factory engines cost £30-40 per horsepower to erect, compared with only around £10 for a waterwheel purchased 'off the shelf'.

In many instances, water power proved to be a cheap and reliable alternative to the steam engine. Indeed, the best water-powered sites (for example, Belper, Darley Dale, Styal and New Lanark) held their own against steam competition until the advent of the electric motor.

Moreover, the best sites on the large rivers could produce considerably more power than early steam engines. William Fairbairn was constructing 200 horsepower (hp) capacity waterwheels in the 1830s, not long after when Joshua Field, a machine tool manufacturer on a trip in 1821, had described a 60hp steam engine as a large one. The average, particularly in the cotton mills, was between 30 and 40hp.

The pattern emerging seems to be this: small establishments already using water power were loathe to switch to a new, expensive and potentially unreliable rotative steam engine when the waterwheel they already had was adequate for the job. The cost savings of retaining water power were such that, if supplementary power was needed, millowners preferred installing a recirculating engine, often of the relatively simple Newcomen type. And it may be that if a really large mill was planned, there was much more of an incentive to use water (provided it was available) than steam. How was this the case? To the answer this, I will return to consider James Watt's opinions on the future of steam power.