Vaclav Smil's Millennium Essay “Horse power” (Nature 405, 125; 2000) came as a pleasant surprise to me, as I would not have imagined that the readers of Nature were interested in this topic.
However, it is not true that “ancient throat-and-girth harnesses choked the animal”. This mistake was first made in the 1920s by Richard Lefebvre des Noëttes. His theory that slavery stemmed from the inability of the ancients to use animal power efficiently was very popular in his day. It lingers on in popular literature, although it has long been disproved: as Jean Spruytte showed in the 1970s, there were several ways of harnessing horses in antiquity, none of them choking the animals, and there is no connection between animal harnessing and slavery. (See, for example, Spruytte's Early Harness Systems, J. A. Allen, London, 1982.) Lefebvre des Noëttes had two different harnessing techniques mixed together in his mind. So Smil is right that the horse-collar was an improvement, but only an improvement on a existing technique.
Second, as far as one can tell, ancient and medieval horses were very small, often barely the size of present-day ponies. There is no incontrovertible evidence of the breeding of “heavy war animals needed to carry armoured knights” that has so often been supposed. In medieval times, large horses were a rare luxury. We lack the data on horse size to know what happened before the eighteenth century, so we cannot know whether armoured knights did ride big horses.
My last point concerns ploughs. The replacement of wood by iron and steel obviously allowed many improvements in the general structure and design of ploughs. But the case of mould-boards is special. The idea that “iron mould-boards only crossed from China to Europe in the seventeenth century” is speculative. There is no evidence of metallic mould-boards coming from China to Europe in time to be used as models by European makers of ploughs. (Chinese mould-boards, incidentally, were made of an alloy, cast iron, rather than of plain iron.) In Europe, wooden mould-boards were simply covered by more and more iron sheets to protect against wear. In some regions, wooden mould-boards were made with a curve from late medieval times.
Finally, wooden mould-boards had their own advantages. In the Gâtinais, north of Orléans, for example, arable soils are quite clayey and stick to iron mould-boards, whereas wooden mould-boards get soaked on their surface, forming a lubricating film of water that prevents the earth from sticking to it. Hence wooden mould-boards were used in this area even when ploughs were made completely of iron, up until the era of the tractor.