Until the invention of the railroad, water transport was much more energy efficient than land transport. A bag of grain in late imperial China rose almost 3 percent in price for every mile it had to be carried overland; a lump of coal 4 percent. So where goods were heavy the cost advantages of water transport could be immense: as late as 1828, some Atlantic seacoast towns in the United States found it cheaper to use English coal for heating than to lug wood from the enormous forests that started just a few miles inland.

Nonetheless, far more ton-miles of goods went by land than by water. Much of this was simple geography: Since the vast majority of production and consumption didn’t take place right next to waterways, almost everything that moved went at least partway by land. Moreover, energy efficiency and economic efficiency were not the same thing. True, an animal carrying a load had to eat, but if there was plenty of grass by the side of the road, this might not cost the shipper anything. And if the animal was going to be on the move in search of grass anyway, even long-distance land transport could be astonishingly inexpensive. Often one didn’t even need to build much of a road—if the land was flat and enough of it uncultivated, the beasts would simply make their own paths as they went. Only where the population was too dense (and land too expensive) for foraging along meandering paths was pre-industrial land transport bound to be painfully expensive—and these were often places where waterways were good. (Both the Netherlands and China’s Yangzi Delta, for instance, despite plenty of money, trade and engineering skills, had dismal road systems; there was simply no way to bring the costs of land transport down to where they could compete with water anyway.)

In Mesoamerica, the absence of waterways and large beasts of burden did not prevent the Maya and Aztecs from moving goods over enormous and astoundingly difficult terrains. Trade traveled thousands of miles on the backs of men. Pack trains of hundreds of tamames (carriers) linked the aristocracy of distant areas. But here it was coerced labor and tributary goods, not commodities produced for profit, which filled the roads. Status and power, not economic calculation of gain and loss, motivated trade.

Whether on land or on water, natural constraints mattered. Except where geography was unusually favorable, it was mostly products with high price-to-bulk ratios that were worth shipping long distances: silks, gold and silver, sugar and medicinal herbs, not wheat, limestone or wood. Thus transport powerfully shaped the geographic division of labor and the nature of demand, even where it was good enough to allow a long-distance division of labor to emerge. Sending bulky rice down the Yangzi River and expensive textiles back up against the current was economically viable; reversing those directions would not have been. Shipping fine swords and linens from Spain through Argentina to Potosí was profitable, but exporting wheat, mules or wine from northern Argentina to Spain was inconceivable.

Transport costs limited the size of cities as well, because bulky goods like food and fuel could only come so far before they became too expensive—unless, as in the exceptional case of Potosí, the lonely city sat atop a mountain of silver, enabling the residents to pay sky-high prices without flinching.

Before the 19th century, maintaining a competitive edge in trade was difficult. Centers of overland commerce such as the cities along China’s famous Silk Road depended upon political peace to ward off the depredations of armies and bandits. Overland trade routes varied with the fortunes of war. Maritime trade advantages were also at risk because the key to cheap shipping was ships. And ships, in turn, needed masts made from large, difficult to transport timbers. From Venice to Xiamen to the Americas, great shipping and trading powers found that they either had to secure increasingly remote waterside sources of big trees or allow others to take over shipbuilding. By the 18th century, South China had many of its big junks built in Southeast Asia; on the eve of the American Revolution, one-third of the British merchant fleet was built in the New World, while the Royal Navy struggled to monopolize potential masts from places as remote as Quebec and Madras. Many of the Portuguese ships that plied the triangular trade route between Europe, Africa and America were built in Bahia, Brazil.

Nature also shaped the rhythms of trade and the places where it was conducted by constraining transportation. All across maritime Asia—from Canton to Mocca—trading schedules were dictated by the monsoon winds. Since strong winds blew consistently in one direction for several months and then stopped, and then blew consistently the other way for months, it made no sense to fight those winds. A trader went as far as he (or occasionally she) could in one direction and then stayed around until the wind reversed; his goods were then picked up by another merchant who had arrived earlier and knew precisely how long into the next season he could safely stay and still have enough days of favorable wind to get home.

Thus, instead of Chinese traders spending two or more monsoon seasons (and years) sailing all the way to, say, Persia with silks, it made more sense to sail out one monsoon season and exchange with intermediaries based in between and thereby return home with frankincense and rugs. A series of emporia developed at sites such as Melaka, Surat and the Muscat that had more to do with how far one could travel from there in one sailing season than with what goods could be produced locally. The result was a remarkably lively and cosmopolitan chain of port cities along the Asian littoral, but in many cases these cities had only weak relationships with their immediate hinterlands.

And despite its remarkable efficiency, the system had certain natural limits that no advances in either seafaring or commercial institutions could exceed in the days before steam. Since no merchant could turn back to home before the wind shifted, there was no way to cut the amount of time away from home (and thus the cost of sustaining the crew away from home, as well as the turnover time for capital) below a certain level. In the Atlantic, by contrast, the wind patterns imposed less severe constraints. Major ports arose either because Spanish mercantilism designated them as monopoly entrepôts such as Havana, Cuba, Veracruz, Mexico and Cartagena, Colombia, or because British laissez faire allowed economics to dictate their growth. In the former case, government fiat rather than winds set the departure time. In the British case, a shipper who could cut his turnaround time in port could turn his capital over faster, and cut his expenditures on wages for his crew as well.

This is precisely what happened in the 18th century as Scottish traders built warehouses, appointed agents to collect goods in advance and found other ways to cut their time in New World ports by several weeks on each trip. The results were dramatic, and not only for the traders themselves. As trans-Atlantic shipping costs fell thanks to these innovations, colonists could move farther inland (thus incurring higher local shipping costs) and still get their tobacco, rice and other goods back to Europe at competitive prices. And since most settlers had cash debts to pay (for passage, start-up costs and industrial goods), it was only when one could successfully export from farther inland that Europeans could begin populating areas farther from the coast—with all that implied for them and the people they displaced.

Kenneth Pomeranz is a professor of history at the University of Chicago. With Steven Topik, he is the author of The World that Trade Created: Society, Culture and the World Economy, 1400 to the Present (M.E. Sharpe).