A couple of weeks ago I was in Trier on the invitation of Prof. Christoph Schaefer to give a paper in a series on ‘Connecting the Ancient World: Mediterranean shipping, maritime networks and their impact’ – a series that I really wish I’d heard the rest of, but will just have to wait for the publication. The real pleasure of the trip, besides the excellent hospitality and the fact that Trier is a wonderful city for all lovers of Roman stuff – I’m already thinking about organising a student trip, to give me an excuse to go back – was the opportunity of hearing about the work of the ancient history team there, and questioning a few of my own preconceptions about methodology.
Experimental archaeology, for example; there’s something about building replica ballistas in order to fire iron bolts at dummies wearing replica armour that seems too much like fun to be proper academic research. I do now feel that I have a better appreciation of how the development of such replicas offers insights into Roman practices and patterns of thought in the development and refinement of technology (though there’s also no denying that it makes for a pretty spectacular display in the department foyer, and doubtless serves to pull in some curious students from other periods of history). I’m even more persuaded of the usefulness of building replicas of different types of ships in order to analyse their performance, especially the ability of ancient sailing ships to sail towards the wind. But who needs actual replicas? That’s not how boats or planes or cars tend to get produced or tested these days. Arne Doepke showed me the work he’s been doing, in collaboration with various Trier science departments, in constructing a virtual model of a Roman ship that can then be evaluated through computer simulations far more effectively than any real vessel (and can even be printed out by a 3-d printer).
Even more impressive – whether because the research is more advanced, or because it relates more directly to specific questions I’m interested in – is the work of Pascal Warnking on Roman trade routes. Now, presumably you’ve already had a chance to play with Stanford’s ORBIS model of the ancient world, which indicates the length (in distance, time and cost) of different routes, highlighting the effects of the weather at different times of year, the differentials between sea, river and land transport, the advantages of state-funded travel and so forth. It’s great for illustrating some key general points, particularly because you get to play with it. However, with all due respect to Walter Scheidel and his team, it does have its limitations, in many cases quite deliberately:
By necessity and design, ORBIS models a simplified version of Roman connectivity. By necessity, given the workload associated with any serious attempt to track down every single Roman road and every navigable river, and especially with the computational burden of simulating discrete outcomes for tens of thousands of often only marginally different sea routes. Much the same is true of the cost of incorporating more detailed wind data or ubiquitous low-velocity surface currents.
The number of data points for wind and currents within the sea areas is therefore extremely limited; rather than allowing sailors within the model to sail on any reasonable route, it’s assumed that they will either have hopped from port to port along the coast or followed a small number of standard routes. For many purposes that’s fine – but it doesn’t actually help if what we’re interested in is the routes themselves, since the (fairly limited) ancient anecdotal information is simply imposed on the world of the model, rather than the model seeking to reconstruct the environmental conditions that shaped the main sailing routes.
Pascal’s research starts with the wonderfully detailed information about wind conditions and currents all round the Med that’s available for sailors – data that includes the likelihood of different strengths and directions of wind at different places at different times of year – and then makes use of the software that ocean racers use to determine the best route in given conditions (calibrated for the assumed performance of ancient vessels in sailing towards the wind – and indeed one of the important results of his research is a clear reason for assuming that they could normally sail only at 70 rather than 60 degrees to the wind; it’s a matter of modelling how frequently a vessel could match the record times recorded for different routes by Pliny the Elder – if ancient ships could sail at 60 degrees to the wind, they would have matched these ‘record’ times far too often for them to be worth recording). The programme runs the simulation for different journeys time and again, so that the most common patterns (reflecting the best response to the prevailing conditions) become clear. Suddenly, for example, the relative connectedness of different locations becomes clear; suddenly the effect of the Mistral on sailing routes in the Gulf of Lion is unmistakable; suddenly the need for oars if you’re trying to get round the Aegean at any speed is rather obvious.
What’s most striking in this work is the capacity of complex software and high-powered computers to address important historical questions – if someone has the imagination to conceive how they might do this, or alternatively to think of ways of setting up the questions so that they are susceptible to such solutions. Maybe you just have to know a bit about ocean racing to have the sudden revelation that, hey, that navigation software could be used to explore ancient trade routes as well. One problem, for those less versed in such tools (i.e. me), is the need to take an enormous amount on trust; in the same way, I suppose, as we cannot re-dig the site that we’re reading about in a report, but can only analyse the stated methods and the results as presented by the excavator, so we’re heavily reliant on the researcher’s own presentation of what’s going on, without being in a position to check it directly. I had certain reservations about ORBIS – not because its authors got anything wrong, but because I disagreed with some aspects of their approach and some of their assumptions – but couldn’t do anything about this beyond feeling sceptical of the results. I feel much more persuaded by Pascal’s results, on first encounter – but actually I’m only marginally better informed about his methods because I could get him to explain them to me, slowly, several times…
There’s also the familiar problem that this represents yet another skill that the young researcher ought to possess: Latin, Greek, archaeology, epigraphy, papyrology, assorted modern languages, general social sciences and statistics, and now computer modelling and programming as well. One of my reactions to this fascinating visit to Trier was that I’ve seen another example of the historiography of the future, and it’s amazingly exciting. Another was to feel very old and somewhat redundant…