I also read a great theory once (but I can't find it again through google!) that the industrial revolution really required the invention of the cannon first. Building an efficient steam engine in roman times would have been incredibly expensive and dangerous, because of their primitive metal technology. The cannon created a tech race between nations to develop more/better/cheaper/lighter cannons. Which resulted in better metal working tech. Once you can make big pieces of metal that can withstand and direct the pressure created from explosions, steam engines aren't too far away.
But Heron's invention was still useful, and would have been the fastest moving thing ever built up to that time. Perhaps it could have been adapted to military applications, like flinging tons of rocks really fast at high speed?
See in particular episode 6 in the original 1978 series (just going by the summary on Wikipedia).
Europe's coal belt starts in Scotland, runs through England, into France (and a nubbin in northern Spain / Portugal), Germany, and Silesia.
Oddly enough, the first countries to industrialise were, in order, England (which wasn't busily falling to pieces and/or fighting off invaders), then (after they'd stopped falling to pieces and/or fighting off invaders) France and Germany. They remain the heart of the European economy, a century and a half later.
The third great set of coal deposits were in the United States, though that was a country that also had tremendous forests, more readily accessed (and frequently local to where energy was needed). Coal consumption didn't exceed wood as a fuel until the 1880s. The US also benefited from oil, and was the world's leading extractor of oil from the 1860s to the 1950s. Having first developed coal, iron, railroads, canals, and steamships, the US could usefully transport that oil from the remote rural regions in which it was found (Titusville, Pennsylvania, Ventura, California, Spindletop, Texas) to the urbanised regions (New York, Philadelphia, Boston) where it was used.
The bootstrapping of technology was itself a really interesting story -- gunpowder, metallurgy, iron, fuels, steel, electricity, and prime movers, most especially diesel and gas turbine engines. See Vaclav Smil (Energy in History) and Manfred Weissenbacher (Sources of Power) for more on this.
(The fourth coal belt is in China.)
Rome and the Mediterranian had a long history of deforestation, going to Egypt and Sumer. The first story, the Epic of Gilgamesh, can be read as a story of ecological limits and deforestation.
Roman heating systems, hypocausts, had voracious fuel appetites. I'm not finding a solid reference but believe one could consume a ton or more of wood daily. That was imported from distant locations.
Early industrial windmills and water wheels typically produced 10 - 20 kW of power, with perhaps a few thousands throughout Europe, for a total power capacity of a few tens of MW. The power availed by coal and oil is thousands and millions of times this.
U.S. woodfuel consumption wasn't sustainable, and hasn't increased from levels of the late 19th century.
Interestingly our roads were built with bicycles in mind:
https://www.theguardian.com/environment/bike-blog/2011/aug/1...
Many similar sources such as:
http://www.citylab.com/commute/2014/12/the-secret-history-of...
In fact, Dunlop invented the pneumatic tire for his son's tricycle and they were initially used for bicycles (https://en.m.wikipedia.org/wiki/John_Boyd_Dunlop#Pneumatic_t...)
As of the late 19th century, horses consumed about 25% of the grain production of the US. Much of that was to provide drayage of people and goods to and from railroad stations or train yards.
A horse was a major capital investment. And a donkey can't carry all that much weight -- a light adult, perhaps, but only just.
The US horse population in 1915 was 20 million, contrasting with about 100 million people. Keep in mind that most of those horses were in commercial use -- drayage, farm traction (a large combine would be drawn by 40 horses), etc.
Numbers for England/UK are harder to come by, but a Google Answers thread gives 3.3 million for late Victorian Britain, and 14 million in Europe in 1800, the latter against a human population of 203 million.
The typical person, or family, owned zero horses.
http://answers.google.com/answers/threadview?id=144565
http://www.geohive.com/earth/his_history1.aspx
Though you raise an interesting question: what was ownership of draught animals throughout time, and how did that compare with bicycle sales in, say, the US, England, France, and Germany from 1880 - 1920 or so. Keeping in mind that mass private automobile ownership wasn't really a thing until the 1950s, even in the US.
Slave labor is kind of an extreme case of the paradox of thrift. Slaves have no disposable capital and they flood the labor market down to zero. Slavery is like a permanent deflationary spiral. There is just no reason to innovate or invest in anything except conquest and luxuries, which describes most ancient cultures well.
Had the Greeks or Romans abolished slavery and set a minimum wage we might be flying to Proxima Centauri in a fusion rocket starship by now.
ps: This is something that distort my perception of the world. Somehow similar to the comma.ai birth, the owner believed the others were caught into a limited view of the world. It's possible that a lot of stuff that we dismiss because we believed in others are "limiting" our existence. But the brownian motion seems unavoidable for now.
It's not so easy to draw firm conclusions from Roman society, because the historical record of that is a little thin. The Nazis employed slave labor, but they had a lot of trouble with sabotage and intentionally shoddy work. Game of Thrones is entirely fiction, and is zero evidence as an "example" of slavery.
From the start of the industrial revolution through about the 1960s, the principle mechanisms were division of labour and application of more energy per worker. There was some automation, but it was largely at a very simple scale -- Jaquard's loom, duplicating patterns on lathes and machining systems, dies (a simple mode for reproducing complex shapes).
But for the most part, what wasn't being substituted for was the worker's skill.
Since the 1960s or so, that's begun changing. Soup-to-nuts factories produce goods with little or no worker intervention. CAD-CAM systems translate design directly to machining.
Automation, in short, replaces the mind of the worker, rather than the muscle.
This also calls to mind the question of just what it is that labour is. I'd consider it a compound factor of production:
1. A prime mover.
2. A complex manipulation system.
3. Skill and processing knowledge.
4. A control system through sensing, processing, feedback, and resulting process management inputs.
Labour-as-muscle has been subject to replacing since the time of draught animals, water, and wind power, though at a vastly greater rate with steam, electrical, and internal-combustion power. The result has largely been an amplification of labour via energy inputs.
Substituting for the information-processing capabilities of labour is another matter. It does mean, however, that increasing amounts of skilled work might be replaceable with a much less-skilled technician. I've seen this trend in the IT world over the past several decades, though there remain points at which greater skills are still in demand.
This makes the viability of coerced labour, including the coerced labour of sweatshops and workhouses, as well as actual prison slave labour, for complex goods, far more viable.
Let that spin your moral compass for a bit.
Not to mention that the parent said "generally".
I read somewhere that one of the problems in antiquity to develop useful machines was the abundance of slave/captive labor. Why do you want to create a complicated inefficient first version of a machine when you have cheap humans being to do the task?