Dr. Judith A. Weller
ROMAN TRACTION SYSTEMS
BACK TO INDEX
LOAD LIMITS

One of the most argued points and for many a defining point in this debate is the issue of load limits – that is to say how heavy a load could a team of horses haul using the dorsal yoke versus the Medieval shoulder collar. Despite exaggerated claims for loads in the Middle Ages, a careful examination of the documentation proves once again that there is no significant difference. In short with the dorsal yoke a team of horses could pull just as heavy a load in Roman times as horses harnessed with a horse collar could in the Middle Ages.

Much has been made of that part of the Theodosian Code and the load limit which it placed on certain types of conveyances. Lefebvre des Noëttes was quite sure that the limits set in that document were due to the poor harness systems1. As a result of his 1910 experiment the Commandant claimed that 500kg. (only slightly more than the 1,500 Roman pounds set for Angaria in the Theodosian code) represented the maximum load which could be hauled by a team of horses harnessed under what Lefebvre des Noëttes believed to be the system of harness2. T. Lynn White uncritically accepted this number and stated "Lefebvre des Noëttes proved experimentally that a team of horses can only pull about 1000 pounds with the yoke harness"3. Without even bothering to use a footnote, Jean Gimpel agreed4.

But let me remind readers that both Gimpel and White accepted Lefebvre des Noëttes' reconstruction of Roman harness as valid and hence the result of his experiment. As I mentioned earlier classical scholars and archaeologists were unwilling to validate the 1910 experiment. In 1977 armed with considerably more information, J. Spruytte with the aid of the archaeologist M. L. Littauer did a series of experiments on various traction systems employed in the ancient world. As I demonstrated earlier Lefebvre des Noëttes committed a major error in his reconstruction of Roman harness. Spruytte's experiment corrected that flaw. Using a correctly modeled Roman dorsal yoke harness, Spruytte demonstrated that a team of horses could comfortably pull 975kg. – nearly one ton5 – double the amount set by Lefebvre des Noëttes in 1910.

Spruytte's number makes more sense when we understand that the Theodosian Code set the maximum ALLOWABLE weight, not the maximum POSSIBLE weight for loads hauled on the State maintained highway system. Furthermore in his study of the Langres wagon, the noted French scholar, M. Molin, in a 1984 work, estimated that the entire weight with driver as about 1 ton6 although he believes the wagon construction could sustain a ton and a half. Michel Polfer, aware of Molin's work, has also studied the Langres wagon and has published the results of his study in a 1988 paper. After detailed calculations for the wagon with a full barrel, he as arrived a weight of not under 1,500kg.7 Therefore based on the above evidence it appears that the actual load limit in Roman Imperial Times seems to have been about 1 to 1 ½ metric ton for one team of equids. Heavier loads probably used tandem hitches.

When it comes to the Middle Ages it is found that exorbitant numbers have been claimed for load limits in this period. The prime example of this egregious inflation of numbers is Jean Gimpel. Let me remind the reader what Gimpel says:

An example of medieval horsepower is given in the building accounts of Troyes. From quarries 50 kilometers away, carters on an average journey drove pairs of horses that pulled wagons weighing 2,500 kilograms and loaded with 2,500 kilograms of stone. When, infrequently, the wagons were loaded with 3,900 kilograms of stone, the horses were pulling 6,400 kilograms. This is a very heavy load in comparison with the maximum load, 500 kilograms, a pair of horses was authorised to pull on roads in Roman times.8

Gimpel is asking readers to believe, unfootnoted, that the carters of Troyes were achieving a load of between 2,500kg. to 3,200kg. per horse, on an average journey of 50km.! But where did he get these numbers? His undocumented source may well have been Jean Hubert's Les Routes du Moyen Age9. Unlike Gimpel this author does believe in footnotes and for a source he turns to the definitive work on supplying stone to the builders of Troyes cathedral which is: Piétresson de Saint-Aubin's article in Bulletin Archaéolgique entitled "La Fourniture de la Pierre sur les grands chantiers Troyens du moyen Age et de la Renaissance". It is to this work I now turn for a thorough analysis of these numbers.

The old cathedral at Troyes had been severely damaged by a fire in 1188 and planning began immediately for the construction of a new cathedral. The starting date for the construction is generally assumed to have been in 1208 with the acquisition of land beyond the old wall10. In 1583 the collapse of the western tower11 initiated a new round on construction on the cathedral. As late as 1588 stones were still being purchased from quarries to continue the construction12. The primary written sources are the building or fabric accounts. It is from these records that Piétresson de Saint-Aubin took his material. It was not until the 17th century before building actually stopped on the cathedral.

Please note the full title and the use of the words "du moyen Age et de la Renaissance" – material in the article will NOT be exclusively from the Middle Ages but will also include the Renaissance. This is important to note as a goodly part of the article deals with documentation from the early 16th century. In fact the work contains very little detailed information about transport in the 12th and 13th centuries. "The earliest accounts are somewhat laconic, offering few details about the work in progress, and the sequence is punctuated by extensive gaps."13.

The quarries were grouped into three main units: 1. the area of Bar-sur-Seine. 2. Tonnerre 3. Barrois all three of which contained hard limestone14. There were also some chalk stone quarries 4-9 km from Troyes whose stones were of poor quality but inexpensive. The article does not concern itself with this stone. The stone from the Bar-sur-Seine area arrived by ship, but this had its own set of problem15. For the stone from the Tonnerre and Barrois quarries, wagon transport was the only means possible16.

The stone of the Tonnerre quarries enjoyed great renown in the Middle Ages and the majority of Troyes churches were constructed with it17. These quarries were situated at a distance of 55-60 km from Troyes (or approximately 34 to 37 miles)18. On the road between Tonnerre and Troyes it was necessary to climb rough hills, then to cross the humid zone of Champagne where heavy wagons sank into the sandy roads, thence across the large prairies of Armance and Landion and into the pot-holes found in the Chaource forest, which itself was cut across by ponds and marshes19. The route must have been treacherous since the Fabric Accounts of 1486-87 detail the loss of wagon filled with stone near the Coussegrey woods20. The carters were recruited from villages located at the mid-point of the journey, and the documents indicate that the journey was done in stages which allowed the drivers to rest at home mid-way in the journey21.

The Barrois quarries were situated even further away – a distance of some 85 km (nearly 53 miles)22. The route over which the carters traveled was even more treacherous than that from Tonnerre. The route cut across the zone of Perthois, full of pools and humid terrain, where the wagon way must have been quite difficult23. Once the carters reached Margerie, or, more likely, Vitry-en-Perthois, they then had an easier drive on wide and undulating roads leading across the vast hills of the chalky Champagne region24.

So how great a load was hauled on a wagon from these quarries? The quantities transported varied a great deal, according to the capacity of the wagons, which were not uniform in size25. Transport was at first paid by the wagon load, but there are no details on the load of each individual wagon26. In 1366-67 the carters were paid by the quartier, which is also not explicitly defined27. The average load was two quartiers28.It was not until 20 years later that payment was made by the cubic foot29.

Once payment was made by the cubic foot it becomes somewhat easier to deduce the actual load carried by carters from the Tonnerre quarries. In 1486-1487 documents indicate that the carters where carrying 3½ cubic feet30. By 1535-36 the Fabric Accounts record loads of 22 cubic feet31. Consequently Piétresson de Saint-Aubin states that the average load was about 15 cubic feet32.

The next issue is how much did the stone from Tonnerre weigh. The only weight estimate is provided in a footnote. The numbers cited are taken from O. Fontaine, Prix élémentaires des matériaux de construction dans L'Aube et à Troyes which appeared in 188033. Fontaine states that the weight of Tonnerre stone varied from about 1,900kg. to 2,400kg. per cubic meter34. Thus 15 cubic feet is a little more than one ton and 20 cubic feet is about 1,400kg.35. Based on the numbers cited above, 1 cubic foot of stone from Tonnerre weighs about 70kg. Also one must note the following – a cubic foot in the Middle Ages was NOT the same as a cubic foot today. In order to calculate the modern cubic meter from the Medieval cubic foot Piétresson de Saint-Aubin used J-B Lachave, Tables de réduction de anciennes measures de l'Aube, 2e edition, which appeared in 183936.

Some loads are mentioned briefly for the Barrois quarries, but the dates for these loads are after 1500 AD. Again the journey was in 2 stages, and I have already mentioned the poor road in the first stage of the journey. On the 2nd half of the journey the carters accepted heavy loads and arrived at Troyes with up to 40 cubic feet per wagon, and never less than 8 cubic feet per wagon ; the same carter often brought 2 wagons on which he sometimes loaded 55 cubic feet37. The documentation for these loads are from the Fabric Accounts of 1503-150438. At no place in the article are there given any weight estimates for the stone of Barrois. It is also noted in the article that the stone from Barrois was not much used until late 15th century and during the 16th century because the cost of transport from Barrois was very high when compared to that of Tonnerre39. If you use the weights from Tonnerre and apply them to the Barrois stone (i.e. 70kg. per cubic foot), you have wagon loads weighing between 560kg. (8 cubic feet per wagon) to 2,800kg. (40 cubic feet per wagon). But using these numbers may not be valid since the weigh of Barrois stone is never given and also the use of Barrois stone is barely documented until the 16th century. Consequently it applies more to the construction done in the Renaissance rather than the Middle Ages.

According to Piétresson de Saint-Aubin, up until the end of the 16th Century "the characteristics of cartages were that it was done in small quantities by numerous carters".40 Thus based on the evidence cited about – the carters were not hauling much more than 1 ton in the Middle ages. Consequently it is difficult to understand the Gimpel numbers. At NO POINT in the article is the weight of the wagons mentioned, yet Hubert, from which it appears Gimpel got his data, cites Piétresson de Saint-Aubin's article as a source. But remember there is no difference in wagon technology between that of the Roman Empire and that of the Middle Ages. Does Hubert assume that the weight of the load had nearly to equal the weight of the wagon? Or perhaps they are extrapolating backwards from Lefebvre des Noëttes statement that some modern wagons (late 19th or early 20th century) had an empty weight of between 500kg. to 1,500kg. while others weighed between 1,500 and 3,000kg.41. Or perhaps they are using Lefebvre des Noëttes assertion that the mean weight of a wagon is between one-third or one-fourth of its maximum load42. But more interestingly Hubert appears to have used the information about the Barrois stone which does not appear until the 16th century, yet he states that the distance was 50 km which implies the Tonnerre stone. That being the case he appears to have used Fontaine's maximum weigh per cubic meter but there is nothing in the documentation which supports applying the higher weight of Tonnerre stone to that of Barrois.

Nor were heavy building materials for churches and cathedrals pulled only by horses. In church construction at Saint Rémy, Laon, Conques en Rouergues tandem hitches of oxen were used43. In fact Hugues, the Master Builder at Conques en Rouergues used 26 pairs of oxen to transport the capitals and bases of the columns from the quarry to the work yard44. Obviously despite the horse collar, the horse was not found suitable for heavy hauling in this case. That's by no means the maximum load moved in the Middle Ages. As further evidence of how heavy loads were pulled in the Middle Ages, there are these interesting facts which further indicate that multiple teams of animals were used to haul heavy loads:

The transport of such stones and of the massive guns that fired them was clearly a major consideration for 14th and 15th century commanders. A Burgundian source of the 1470s says that a large bombard required 24 horses to draw it, a courtaut (crappaude) 8 horses, a medium-sized serpentine or a mortar 4 horses and even a small serpentine 2 horses. In 1388 a single German bombard belonging to the city of Nuremburg required for its transport 12 horses to draw the barrel, 10, 4 and 6 more respectively to draw the wagons containing the tiller, winch and hoarding, another 20 horses to draw ammunition wagons each containing three 560 lb. stone balls and the appropriate gunpowder, a horse for the master gunner, and a final wagon for his 6 assistants and their various tools. Similarly in 1477 two Italian bombards of no exceptional size required a support train of 48 wagons, each drawn by 2 or 4 horses, to transport their tillers, gunpowder, stone shot, quarrels and other equipment, So great was the weight of some artillery, in fact, that roads and in particular bridges frequently required reinforcement in order to take them. In 1453, for example, Philip the Good of Burgundy had to get a 17-foot bombard weighing 7,764 lbs. from Mons to Lille, which involved strengthening every bridge en route with iron supports. When at one point this monster slid into a ditch it took two whole days to get it back on to the road. It is therefore easy to understand why guns and ammunition were frequently transported by river instead, as they were by the English in Normandy and Gascony in the 1420s and by the Burgundians in Flanders in 1453.45
Even early in the 17th century, Surirey de Saint Remy indicated that in the artillery the load of a 2-wheeled vehicle hitched to 4 horses harnessed in file varied from 537kg. to 635kg.46. Even in the 18th century the maximum allowable weight given for a team of horses pulling a heavy wagon on a hard road was 2 tons47, only 1 additional ton higher than that estimated for the Langres Wagon. Dorian Gerhold also states that in the 1700's that the average load per horse was 300kg48 and that by 1800 it had increased to only 550kg.49.

Nor must we forget that the weight of the animal is a factor in how much he can pull. The effort of traction of an animal is approximately 10% of its weight; but for the horse it is slightly more – between 15% to 20% of its weight50. The massive breeds of draught horses such as the Percheron do not make their appearance until the 17th and 18 thcenturies. There is no evidence indicating the existence of large, heavy horses being bred for hauling in the Middle Ages.

Consequently one must agree with John Langdon who states "in terms of loads that could be hauled, there was little difference between the effectiveness of the transport in Roman Gaul and that which came later in the medieval period,"51. We also must concur with Langdon when he writes "a ton was a very respectable load for horse-hauled transport in the medieval period"52.

This leads to the conclusion that after a thorough examination of article by Piétresson de Saint-Aubin that the claims of these Gimpel and Hubert were based more on wishes than on fact.


  1. Lefebvre des Noëttes, 1924, pp. 67-70
  2. Ibid., pp. 85-86
  3. White, p. 60
  4. Gimpel, p. 33
  5. Spruytte, p. 98
  6. Molin, p. 106 – there is some disagreement over the amount of liquid the barrel itself would contain which could adjust the overall weight slightly. In arriving at his conclusion M. Molin has studied numerous other monuments depicting a similar wagon.
  7. Polfer, passim
  8. Gimpel, p. 32
  9. Huber, pp. 39-40 "En moyenne, on chargeait 2300 kilos de pierre de taille sur chaque voiture. La charge pouvait atteindre 3900 kilos. Si l'on ajoute a ce chiffre la tare du char lui-meme, soit environ 2500 kilos, on obtient un total de 6400 kilos, poids qui est sensiblement celui des gros charrois qui circulaient sur nos routes de France au debut du XXe siecle".
  10. Murray, p. 12
  11. Ibid., p. 109
  12. Piétresson de Saint-Aubin, p. 599
  13. Murray, p. 2
  14. Piétresson de Saint-Aubin, pp. 570-571
  15. Ibid., pp. 591-592
  16. Ibid., p. 592
  17. Ibid., p. 571
  18. Ibid., p 571 fn. 4 in which he cites the 1777 work by Milony, Essais sur la bâtisse de Troyes, as the source for this mean distance
  19. Ibid., p. 594
  20. Ibid., p. 594. footnote 3
  21. Ibid., p. 593
  22. Ibid., p. 571 fn. 11. Once again Milony's 1777 work is the source for this distance
  23. Ibid., p. 594
  24. Ibid., pp. 594-595. The authority for this description of the route is Boutiot, Études sur les voies romaines de l'Aube, Mèm. de la Soc. Acad. de l'Aube, 1862
  25. Ibid., p. 594
  26. Ibid., p. 595. in footnote 3, the records cited are from the period 1298-1299
  27. Ibid., p. 595
  28. Ibid., p. 595 footnote 4
  29. Ibid., p. 595
  30. Ibid., p. 594 footnote 5
  31. Ibid., see also footnote 5
  32. Ibid.,
  33. Ibid., p. 570 The full citation of the work is given in footnote 2
  34. Ibid., p. 594, footnote 5. However there is no information given in the records to indicate the density of the stone found in the various loads
  35. Ibid., p. 594, footnote 5
  36. Ibid., p. 598 fn. 1. In constructing the table Piétresson de Saint-Aubin uses the number 0.034 as the multiplier when converting from Medieval cubic feet to modern cubic meters
  37. Ibid., p. 595
  38. Ibid., p. 595 footnote 2
  39. Ibid., pp. 595-596
  40. Ibid., p. 593
  41. Lefebvre des Noëttes, 1924, p. 108
  42. Ibid.
  43. Lefebvre des Noëttes, 1924, p. 97. The source cited is: Victor Mortet, Recueil de textes relatifs à l'histoire de l'architecture, Paris. 1911
  44. Ibid. also cited by Landgon, B&T, FN 46 on page 134
  45. Heath p. 160
  46. As cited by Spruytte, p 108, the 2nd edition was published in 1745, the 1st edition in 1697
  47. Bagwell, p. 13
  48. Gerhold, p. 26
  49. Ibid., p. 65
  50. Abeele, p. 18.
  51. Langdon, Brancards et Transport, p. 116
  52. Ibid.


PREVIOUS SECTION: WAGON CONSTRUCTION | NEXT SECTION: THE HORSE IN AGRICULTURE