This is a bit of a late answer, but your question had wondering exactly how much of William's logistics we could reconstruct. For 14th century England we have an enormous amount of sources on the purchase, processing or manufacture and transport of food and other items, and a lot of work has been done to map out all the logistical elements. Mid-11th century Normandy, though? Not so much. Apart from some charters and chronicles, nothing of what was obviously a relatively sophisticated system of administration remains from the period. Fortunately, there are some amazing scholars out there who have worked wonders with the limited material and, as a result, I think this question can be answered with only limited reference to later medieval history.

Part 1: Working out the Requirements

Let's set the stage a little. In 1066, Normandy was one of the most politically unified and wealthy parts of Northern France. While it wasn't as large as the County of Toulouse or the Duchy of Aquitaine, the Norman dukes had spent a hundred and fifty years securing their hold on their nominal territory, bringing their lords under control through a combination of marriages, diplomacy and military force so that, unlike the lands of the Capetian kings, the Norman dukes were the masters of their own land. William had had to fight off some rebellious vassals in order to stay in power in his youth, and by no means did he have total power, but William hadn't faced any rebellious vassals in over a decade and enjoyed very broad support at home.

Normandy's wealth came from a range of sources, including maritime trade and agriculture, and was well known for its wealth by contemporaries. The fertile plain around Caen allowed food to be exported - primarily to Rouen according to the limited evidence we have, but possibly also elsewhere - while the hybrid grazing/cropping system used in the Orne valley allowed for significant production of sheep, pigs and meat cattle. Although the earliest fulling mill in the region (which is the earliest north of the Alps) is only attested in 1087, there was also no doubt a growing textile trade at the time.

Rouen, the capital of Normandy, had long been the place where Scandinavian raiders or conquerors sold their loot, thanks to the tight connections between the Norman aristocracy and Scandinavia, and it was also the first major city on the Seine. The population, likely 20-25 000 and similar to London, was significant by the standards of Western Europe and this also helped promote trade. In fact, Normandy was relatively urbanised in the 11th century, with Caen having a population of close to 10 000 people - Dieppe, Eu and Fécamp being not far behind - and another five (Verneuil, Bayeux, Pont-Audemer, Falaise and Neufchatel) having a population of 3-5000. How many smaller towns existed is hard to know, as many were only "officially" founded in the period between 1066 and 1150, but England only had three towns of more than 7000 people (London, Winchester and York) compared to Normandy's six, despite Normandy having only a quarter the area of England^1 . Urban trade and production must necessarily have factored into Normandy's wealth.

When it comes to the invasion of England, it's important to consider that William didn't just have to feed the men he brought to Hastings; he also had to feed the crews of his ships, the men he left behind as garrisons in Pevensey and Hastings and also all the non-combatants who built his ships, cooked food or performed any of the hundreds of little tasks necessary for an army to function. Bernard S. Bachrach has argued that William had 14 000 men at Dives for the month before he ultimately sailed - 10 000 combatants and 4000 non-combatants - but here I think even the well known maximalist of medieval military history has underestimated the logistical magnitude.

Estimating the numbers of any army in this period runs into many of the same problems as ordinary demography, but in this case we have two important advantages. The first is that we have the Skuldelev 1, a Danish cargo ship from 1030 that was sunk in the Roskilde fjord as part of a temporary defence and which was largely preserved. This gives us a good example of the most common type and size of cargo ship in use in Normandy, England and Scandinavia, and a replica gives us an idea of its performance and the minimum crew needed.

The second is the Infendationes Militum, a list of military obligations that the Norman lords owed to Henry II in 1172 in addition to the total number of knights in their service. The Bayeux Inquest of 1133 shows that the bishop of Bayeux, at least, owed the same number of knights in 1133 as in 1172, and the Ship List of William the Conqueror, which survives in what is now considered a 12th century copy of an authentic 11th century document, confirm that the Lords of Montfort and Giffart had agreed to supply as many knights in 1066 as they had enfeoffed in 1172. While we can't be entirely certain, it seems likely that there were as many knights available in 1066 as in 1172, and this gives us a starting point.

Christopher Hewitt, in his PhD thesis and an earlier article, has made use of GIS software and a list of every individual even remotely associated with the battle, to plot out where each of the named Norman lords came from and, using the Infendationes Militum, estimate the number of knights who accompanied William. He arrives at a figure of 2000 cavalry, although it's not entirely clear whether he is factoring in the foreign mercenaries.

The Infendationes Militum only lists 1846 knights, but only 75% of the Norman lords returned an answer and it's estimated that the full number was in the region of 2500, so the figure of 2000 seems to solely for the Normans. The Bretons feature prominently in most contemporary accounts, along with a long list of other regions of France and Italy that the mercenaries came from, making up the left wing of the army. One of these, Viscount Aimeri of Thouars, was a powerful lord from Aquitaine and likely William's first cousin by marriage, and could easily have brought more than a hundred knights with him. On the whole, factoring in the need to keep some knights behind to garrison important border castles, and the left wing being entirely made up of non-Normans, I believe that 2500 knights is a more reasonable figure.

This instantly gives us some more numbers, as we can assume a minimum of 5000 horses and 2500 squires/servants. Although every author I've read has assumed that each knight only had one horse, the 1101 treaty between the Count of Flanders and Henry I assumed that each knight would have three horses to be transported, so I think it's far more likely that William's knights had at least two horses each, just as later Crusaders traveling long distances did. Each knight would also likely have been accompanied by a squire or a servant, some with multiple of each, because during the Battle of Hastings Bishop Odo encouraged them and possibly led them in a counterattack after the initial Norman rout. The number of infantry is a mere guess, but based on documented 13th and 14th century numbers, the proportion was probably 3:1 infantry to cavalry, so c.7500 altogether.

5000 horses and 12 500 men is a very large force by medieval standards, and would have required a large fleet. Light cargo ships like the Skuldelev 1 could have carried six horses and their riders in some discomfort, but there were also larger ships which may have resembled the late-12th century "Big Ship" from Bryggen that could take twenty to twenty five horses and forty to fifty men^2 . Bacharach has argued that William constructed Mediterranean style horse transports, based on Wace's description of the Normans riding off their ships fully armed in addition to some other hints of Italian or Byzantine seamanship, but even if William wasn't constructing this specific type of ship it would make sense that the ships he built were orientated towards horse capacity.

Estimates for the total number of ships, based on the Ship List and a reference in William of Poitiers to the number of ships that a Latin summary of the Iliad claimed went to Troy, come in at around a thousand ships total. If 100 of these were large enough to carry an average of 20 horses each, 430 Skuldelev class vessels would be required to carry the other horses. The remaining men could be carried in 300 Skuldelev style vessels, leaving 170 vessels to carry cargo, for over 4000 tonnes of cargo capacity. These ships would require a total crew of ~7000 sailors, meaning that William had to feed twenty thousand men at his own expense for the month at Dives^3 .

Twenty thousand men and five thousand horses will require 20 tonnes of unmilled grain for bread, 40 000 litres of wine, 2 tonnes of dried legumes, 2 tonnes of meat, 1 tonne of cheese, 17.5 tonnes of oats and 35 tonnes of hay, each and every day^4 . Additionally, the grain had to be milled into flour and baked into bread, the horses had to be watered and the meat cooked or put into pottage. In round terms, including the weight of the barrels, that's 138 tonnes of cargo each and every day. As the only period where we have any indication of the number of men and animals being fed is the month the whole army was at Dives, I'll be concentrating on that.