Napoleon's Army Was Starving Before Its First Big Battle

The Grande Armée crossed the Niemen on 24 June 1812. By the battle of Smolensk on 16 to 18 August, the main body had lost about one third of its men, and Wikipedia's summary says this happened before any major battle was fought [1]. Hunger, disease and the need to garrison depots caused most of it. Russian tactics caused little of it. I want to test whether the wagon-train design explains the timing. My answer is: probably yes, but open sources do not prove it, and I must say first what I could not do.

The question

Did the supply system set the date when the army began to fall apart? Or did the decline come from some other cause, such as weather, disease or Russian action?

My working thesis said that published tables by Chandler, Riehn and Nafziger show this. I did not open those books. I read secondary accounts that cite them: Wikipedia's article on Smolensk cites Nafziger (1988), Chandler (1996), Bodart (1908) and Mikaberidze (2007, 2016) [2]. So I cannot say what the tables themselves print. I can only say what the accounts that use them report. That is a weaker claim, and it is the claim I defend here.

I also did not run any code. Every calculation below is simple arithmetic that a reader can repeat.

Data and where it came from

Force numbers for 1812 are quoted with no date or source more often than for almost any other campaign. Source and date? Here is what I found, with the spread left visible.

Item	Figure	Source
Strength at the Niemen	"exceeded 450,000", with over 150,000 horses and about 25,000 wagons	Wikipedia, French invasion of Russia [1]
Strength at the Niemen	615,000 total, 200,000 horses, 7,848 supply vehicles	World History Encyclopedia [3]
Strength at the Niemen	about 685,000 on 24 June 1812	Napoleonic Archive [4]
Main force before Smolensk	375,000 reduced to 185,000 within five weeks	Wikipedia, Battle of Smolensk [2]
Train	"twenty train battalions with 7,848 vehicles", a 40-day supply	Wikipedia, French invasion of Russia [1]
Train	"Twenty six transport battalions", about 25,000 wagons, 200,000 animals	historyofwar.org [5]
Vilnius depot	rations for 100,000 men for 40 days	Wikipedia, French invasion of Russia [1]
Halts	Vilnius 4 days (late June), Vitebsk 15 days (late July), Smolensk 7 days (late August)	H. G. W. Davie [6]
Baggage allowance	12.5 wagons per 1,000 men	H. G. W. Davie [6]

These numbers do not agree. The Niemen strength ranges from 450,000 to 685,000. The train ranges from 7,848 vehicles to about 25,000 wagons. The two vehicle counts probably describe different things. One may count only the military train, and the other may count all wagons with the army, including regimental and requisitioned ones. I do not know which, and none of my sources says. The figure of 7,848 vehicles also appears in two sources that may copy a common reference [1][3].

One more discrepancy matters. Wikipedia's Smolensk article speaks of a 375,000-man main striking force reduced to 185,000 [2]. The same article gives about 180,000 troops with Napoleon in total, with 45,000 engaged at Smolensk itself [2]. Historyofwar.org gives about 156,000 at Vitebsk [5]. These are different units, different dates and different definitions of "the army". Nobody can compare them without stating the base. My open sources do not always state it.

Method

The method is a timeline test with three steps.

1. Count the days from the crossing to each known halt or battle.
2. Compare those days with the stated supply endurance of the train: 40 days.
3. Check whether the reported fall in strength happens before or after the date when the train should have emptied.

This is a plausibility test, not a model. It cannot show cause. It can show whether the timing is consistent with the supply explanation.

The count of days, from 24 June 1812:

$d_{\text{Vilna}} = 4, \quad d_{\text{supply}} = 40, \quad d_{\text{Smolensk}} = 53$

Day 4 is 28 June, when Napoleon entered Vilna [5]. Day 40 is 3 August. Day 53 is 16 August, the first day of the battle of Smolensk [2]. Day 56, the end of eight weeks, is 19 August. The Vitebsk halt was in late July, roughly day 34 by my count, but my source says only "late July" [6], so I treat that as approximate.

Result

First result: the nominal 40-day supply would run out on about 3 August, if the whole army drew on it from the first day. That date falls between the Vitebsk halt and Smolensk. It falls before any major battle. It also falls before the eight-week mark at which the one-third loss is reported [1].

Second result: the nominal date is probably too late. Three sources say the train could not keep pace with the advance. Davie describes a wagon convoy working as a rolling magazine [6]. One source states that "no supply wagon could keep up with a forced marched infantry column" [5]. The Smolensk article says rapid forced marches and the failure of wagon trains led to desertion and tens of thousands of losses to hunger and dysentery [2]. The same article says huge numbers of cavalry horses, transport horses and oxen died because of a lack of grazing and because wagons could not carry enough fodder [2].

So the earliest reported damage is not at day 40. The sources speak of losses by the Vitebsk halt. One account gives about 100,000 men missing, mainly stragglers or sick, by that pause [5]. Another says over 80,000 men were dead or seriously ill by the third week of July [3]. Those figures place the visible loss at about day 25 to 35, before the nominal supply date. In my reading, that is the strongest evidence for the thesis. The train failed at the front before its stock ran out in the rear. The Vilnius depot held rations for 100,000 men for 40 days [1], which is a fraction of the force that crossed. The depot could not feed the whole army.

Van Creveld's argument in Supplying War (1977) points the same way, according to the summary I read: low population density left little to seize, and with distance the draught animals ate more than they could carry [7]. I did not read the book itself, and the summary is a blog post. Treat it as a pointer, not as evidence.

Arithmetic on the loss shares, with the base named each time:

• If the base is 375,000 and the end figure is 185,000, the loss is 190,000, or about 51 percent. But the Smolensk article says part of the fall came from detachments (90,000 under Oudinot) and from garrisons [2]. Those men were not dead. So 51 percent is an upper bound for loss to supply, disease and desertion.
• A loss of "a third in eight weeks" [1] gives about 150,000 men on a base of 450,000 and about 228,000 on a base of 685,000 [4]. The two results differ by about 52 percent of the lower one. That is far wider than my stated 15 percent confidence band for open-source force numbers. The band is too tight for 1812.

I do not claim that these figures are the real loss. I claim that the range is a fair summary of what the open sources say.

Sensitivity: which assumption moves the result most

Three assumptions matter. I rank them by how much each moves the conclusion.

1. The base. This moves the result most. The answer to "what share was lost" changes from about one third to about one half depending on whether the base is the whole invasion force, the main force or the force with Napoleon. Detachments and garrisons count as losses in some sources and not in others [2]. If the reader changes the base, the size of the claim "supply caused most of it" changes by a factor close to two. I keep the direction of the claim but I do not give it a number.

2. The train size. If the train was 7,848 vehicles, the 40-day stock was thin for a force of 450,000 or more. If it was about 25,000 wagons [5], the stock was larger, but the sources I read do not say how much was food and how much was ammunition, ovens and baggage [6]. The date of exhaustion moves by weeks, not days. I did not compute it, because I lack a ration weight and a load per vehicle from a source I could cite. Any figure I gave would be my own guess in the form of a number.

3. The cause. Disease (typhus, dysentery, diphtheria) is named beside supply shortage in the Wikipedia summary [1]. Hunger and disease are linked, because weak men fall ill. I cannot separate them with open sources. Whoever wants to credit Russian action will find it in the fighting at Smolensk, with 10,000 French casualties by Chandler's estimate [2]. That is small beside a loss of 150,000 or more before it.

The strongest objection to my reading is this: the wagon design may not be the cause, and the cause may be Napoleon's choice of a fast, wide advance. A train that fails on a forced march might succeed on a slower one. Van Creveld's point is that the system suited a populated Europe, and that is a doctrine and planning issue as much as a design issue [7]. I think the objection has force. The train and the plan failed together, and the sources do not let me separate them.

What the evidence does and does not show

It shows a consistent picture. The army lost about one third of its strength in eight weeks [1]. The loss was reported by late July [3][5]. The smaller battle at Smolensk came after the damage [2].

It does not show that the wagon design set the date. For that I would need the original strength returns that Chandler, Riehn and Nafziger use, a load table per vehicle type, and a ration weight, all with dates. I have none of these in hand. Anyone who quotes a single figure for the Grande Armée in 1812 without naming a base, a date and a source is quoting a number that this evidence cannot support. The supply officer would like a word.

I will correct this post if the tables show a different timing. A strength return that shows no large loss before about 3 August, for example, would weaken the thesis.