The age of steam in ocean shipping dates from the middle of the 19th century and the transition from sail to steam has been credited with a massive expansion in international trade (Pascali 2017). But it has long been recognised that sail continued its ascendency at least until the 1880s (Graham 1956). Steam first displaced sail on shorter routes and then gradually diffused to longer transoceanic voyages as steamships became more efficient, particularly in their use of coal (Harley 1989). Indeed, this era has often been seen as one in which an advancing technology gradually ousted a relatively stagnant one. Yet one reason that sail proved to be so resilient is that sailing-ship technology also continued to improve. Up to the middle of the 19th century, advances in design and construction methods, and above all the copper sheathing of hulls, contributed to increases in speed under sail (Kelly and Ó Gráda 2018). But, with the advent of the clipper ships and the replacement of wooden hulls by iron, the best was yet to come.
A classic example of the resilience of sail is on the route from Europe to Australia where sail dominated the outward transport of emigrants up to the early 1880s and the return shipping of bulk goods such as wool until the turn of the century. A key measure of efficiency is speed. The duration of emigrant voyages to Australia under sail fell from 124 days in 1837-41 to 86 days in 1879-83 (Hatton 2023). Much attention has been paid to the majestic clippers and subsequent increases in the size and robustness of sailing ships in accounting for the falling voyage times. But another factor is the routes they travelled through the Atlantic and Indian Oceans. From the 1840s sailing ships rarely stopped at the Cape of Good Hope and, inspired by research into winds and currents by Matthew Maury and John Towson, they increasingly followed the so-called great circle route. But to date there has been no quantitative analysis of how the quality of ships affected their voyage durations. Particularly intriguing is the possible complementarity between the quality of ships and the routes that they followed through the hazardous southern oceans.
In a recent study (Hatton 2024), I analyse two datasets of emigrant voyages to Australia. The first is for 311 voyages of ships that were chartered to take assisted emigrants from the UK to Adelaide, South Australia. As Figure 1 shows, from 1848 to 1885 the average duration of these voyages decreased by 25 days, or 5.6 days per decade. They were sailed by 232 different ships and so there is a wide range of variation and each voyage has been linked to the key characteristics of the ship. These include the age of the ship, its gross tonnage (a measure of capacity, not weight), its rigging (whether as a barque or a square-rigged ship) and whether it was built of wood or iron. Finally, while clipper-style ships were streamlined in shape and designed for speed there is no precise definition. In the absence of a better definition, I designate a ‘clipper’ as where the length of the ship is at least five times its breadth.
Figure 1 Average durations of migrant voyages from the UK to Adelaide, 1848-1885
Source: See Hatton (2024).
Analysing these data, I find that larger ships and iron ships were faster – the latter by about five days. Barques were slower by about three days, which makes sense as barques (where the rearmost sail was rigged fore and aft rather than square) were more manoeuvrable but slower in following winds. But these effects weaken when the variable for ‘clipper’ is also included; voyage times were more than a week shorter for ‘clippers’. Taken together, changes in the attributes of ships can account for almost all of the decline in voyage times shown in Figure 1. That seems to leave little room for any contribution for improved navigation to reduce voyage durations. However, it is possible that ships with superior sailing characteristics were faster partly because they were better able to exploit faster nautical tracks, especially in the southern oceans. For this it is necessary to examine the tracks that ships actually followed.
To investigate this issue, I use a unique dataset of the tracks of 290 sailing ships that arrived in Melbourne from Europe in 1854-62. These were originally collected from ships’ logs by George Neumayer, director of the Melbourne observatory and they are mapped in Figure 2. On the leg running from the English Channel into the southern Atlantic, the dots represent the average position of ships’ longitude for benchmarks of latitude, and the whiskers are one standard deviation either side of the mean. It shows that these ships typically followed the trade winds along the west coast of Africa before sailing through the equatorial doldrums towards the eastern tip of Brazil and then looping south and east. From the prime meridian to 140 degrees east the dots are mean latitudes for benchmarks of longitude, again with whiskers of one standard deviation. This shows that the average track was a few degrees further south than 40 degrees, which enabled ships to exploit the strong westerlies of the roaring forties.
Figure 2 Tracks of voyages from Europe to Melbourne
Source: See Hatton (2024).
I use these data to assess the links between key elements in ship’s tracks and their voyage times. On the southward passage, the fastest passages are by ships sailing east of Cape Verde but keeping away from the African coast. On the eastward leg the key to faster sailing is steering further south – as high as 50 degrees south. This of course was a hazardous route where ships could encounter severe storms and mountainous seas, not to mention icebergs. I also find that faster passages to Melbourne are linked with ship characteristics, specifically barque-rigging, ship tonnage and ‘clipper’ design, in the same way as they were for voyages to Adelaide.
In order to account for the interplay between the qualities of ships and the tracks that they followed. I turn to mediation analysis. This is illustrated in Figure 3. It shows that a given characteristic could be linked to voyage duration either directly, or mediated via the track navigated, or through a combination of both channels. On the southward leg, tonnage is mainly a direct negative association with duration, while barque is positive and works mainly through mediation (i.e. not sailing east of Cape Verde). The negative ‘clipper’ effect is partially mediated (about 18% of the total effect). On the eastward leg, tonnage and barque are mainly direct effects (negative and positive respectively) while ‘clipper’ is partially mediated with both direct and indirect effects. Indeed, 43% of shorter durations for ‘clippers’ is mediated through taking a more southerly ‘great circle’ track between the prime meridian and 140 degrees east.
Figure 3 Mediation analysis
Conclusion
Steam ultimately triumphed over sail, but it took decades for that triumph to be completed, partly because sail proved to be so resilient on the longer routes. While that story is fairly well known, this study shows that the increase in speed under sail can be linked to specific characteristics of sailing ships. Less often recognised is that, although larger and more streamlined ships were faster on a given track, they were also better able to exploit faster tracks. Over the whole voyage from Europe to Australia about one third of the advantage of ‘clippers’ over other ships was linked with sailing the so-called great circle route. It should come as no surprise that this became known as the clipper route.
References
Graham, G S (1956), “The Ascendancy of the Sailing Ship 1850-85,” Economic History Review 9 (1): 74-88.
Harley, C K (1988), “Ocean Freight Rates and Productivity, 1740–1913: The Primacy of Mechanical Invention Reaffirmed,” Journal of Economic History 48(4): 851–876.
Hatton, T J (2023), “Voyage Durations in the Age of Mass Migration,” VoxEU.org, 31 July.
Hatton, T J (2024), “Sailing Ship Technology, Navigation and the Duration of Voyages to Australia, 1848-85,” CEPR Discussion Paper 19342.
Ó Gráda, C and M Kelly (2018), “Speed under sail during the early Industrial Revolution,” VoxEU.org, 27 January.
Pascali, L (2017), “The Wind of Change: Maritime Technology, Trade, and Economic Development,” American Economic Review 107 (9): 2821–2854.
