Nikephorian Questions

by Graeme Walker

The book, Sowing the Dragons Teeth, by Eric McGeer contains copies of two similar military manuals, the Praecepta, by Nikephoros Phokas, c. 969, and the Taktika by Nikephoros Ouranos, c.1000AD. The latter often copies the former, word for word, but incorporates changes made during the intervening 30 years, as well as other materials. The work by Nikephoros Phokas apparently borrowed heavily from a slightly earlier work, the Sylloge Taktikorum. The wealth of manuals and commentaries from the 10th century allows us to look at the army organisation of the Byzantine empire in great detail. The current book raises several questions, three of which I would like to discuss here.

1. The correct size of Nikephoros Phokas' infantry army?

At the beginning of his work, Phokas gives the exact number of recruits required for his ideal army, being 11,200 hoplitai (heavy infantry) and 4,800 toxotai (archers). They are to be divided into 12 Chiliarchies (battalions), of 1,000 men each. Immediately there seems to be a contradiction, as only 12,000 not 16,000, men will be required to fill the army.

Eric McGeer, the translator, tried to explain the discrepancy by suggesting that the leftover 4,000 either became reserves to make up for campaign losses, or were kept apart for other duties1 , but neither practice is hinted at within the treatise and reserves do not fit into the defensive role which the foot soldiers were expected to play. Under the circumstances, I think there is a better explanation.

We can begin by analysing the make up of a Chiliarchy, finding out what troops are required, and comparing this with the different troop types available. Each Chiliarchy is made up of 4 distinctive units, these being:

A Chiliarchy

400 pike bearers (hoplitai)
300 archers
200 javelineers
(may include some slingers and archers), on the wings
100 menavlion2 bearers
as a roving unit

1,000 men

When multiplied by 12, to fill the 12 Chiliarchies recommended in the text, we get the total number of men required to fill the specialist units, as shown in the next table (below). If we were to increase the number of Chiliarchies to 16, by adding another 4,000 men, we could expect the figures for troop types to increase as shown in the table after that:

12 Chiliarchy model

4,800 pike bearers
3,600 archers
3,200 javelineers etc on the wings
1,200 menavlion bearers

12,000 men

16 Chiliarchy model

6,400 pike bearers
4,800 archers
2,400 javelineers etc on the wings
1,600 menavlion bearers

16,000 men

Lets consider the figures from the 16 Chiliarchy model first. The menavlatoi and pike bearers can be identified as distinct forms of heavy infantry, and the archers are readily distinguished. If we then assume that the troops guarding the wings, or the intervals between Chiliarchies, come from the pool of heavy infantry, re-armed for special duties, we get the total of 11,200 heavy infantry and 4,800 archers. This formula seems to suit the figures quoted at the start of Phokas' Praecepta, and it can be tested against the Taktika of Ouranos. This later manual also tells us that an infantry army should be divided into 12 Chiliarchies, and Ouranos' text book figures of 6 - 8,000 heavy infantry and 4,000 archers, though less precise than those of Phokas, much better suit the 12 Chiliarchy model in the left column.

Phokas has listed 16,000 troops overall, but they cannot fit into 12 groups of 1,000. McGeer's suggestion that the excess made up reserves ignores the reason why Phokas was so specific about his unit numbers in the first place. McGeer suggests that 333 left over troops were attached to each 1,000 man unit, or were united as a group of 4,000 elsewhere. If so much overlap was available Phokas could have easily given more general figures, such as 11,000 and 5,000 without upsetting textbook calculations. He must have had a reason for being so precise.

I suspect that, in practice, generals tried to make their armies as large as was possible, but that in text books, writers worked with a standard figure of 16,000 soldiers. That a textbook infantry army could be made up of 16 Chiliarchies is shown by McGeer3. He quotes from the manual De re militari4 (c. 995) which recommends an infantry army of 16, not 12, Chiliarchies. This is for use in the Balkans, in the late 10th century, where the terrain restricted the use of cavalry. At some time prior to the writing of the Praecepta there was a change from 16 infantry units to 12. This may have been because more soldiers were being armed as cavalry for the campaigns of conquest in the open terrain around Cilicia and Antioch during the mid 10th century. There are complaints about Phokas, as emperor, forcing many citizens into higher income assessment brackets to finance the campaigns. The same is true for the military, with the critic Zonaras saying that all troops were forced to serve at the next highest troop type. If this is even partly true, there may have been a drop in the number of men available for infantry, as more were forced into the tax bracket from which the cavalry were drawn, and thus a need for revision of the typical army organisation.

If Phokas made a mental slip and used the old overall numbers for combatants, which are not mentioned again, but correctly divided the army into the new 12 part organisation that was in everyday use, we should be able to conclude that the infantry army in his book numbers 12,000 soldiers. He refers several times, to the 12 Chiliarchies and the fact that they are made up of 1,000 men each.

2. How were the Chiliarchies and Dekads organised ?

In the next question I will use a convention. A rank or row of soldiers is meant to indicate a line of soldiers standing one beside another. A file or column of soldiers is a line of soldiers one standing behind another.

While the manuals create a comprehensive picture of the organisation in the ranks, it is not complete and we are forced to deduce some of the information. Classical history students might assume that a Dekad was a file of soldiers measured from the front rank of the battle line to the rear, and that it consisted of 4 pike bearers and 3 archers. I hope to show that, though this looks like an attractive solution, it is not correct.

A Chiliarchy, showing the relative positions of the Hekakontarchs and the Pentakontarchs.

As mentioned earlier, each Chiliarchy is made up of 4 specialist units. The majority of these, the pike bearers and the archers, form their battle line in a block 100 men wide by 7 deep. The javelineers form the wings of this block, and the menavlatoi stand in a row behind, in front or where ever they are needed.

Within the block of 700, the positions of the higher officers are known. Nikephoros says that a Hekatontarch stands in the middle of each line of 100, while a Pentekontarch stands on each wing of the line, and is in charge of the men between himself and the Hekatontarch. Of the officers, only the position of the Dekadarchs is not described. McGeer seems to have thought of the Dekads in the classical model, as files with all the Dekadarchs in the front rank facing the enemy, and with their subordinates behind them. Although he admits that there doesnt appear to be an Ouragos, or file closer, at the rear of each file, he does not consider the possibility that the Dekads were organised in some other way, probably because there is no precedent. The problem with his interpretation is the break in the chain of command. The Hekatontarch and Pentekontarchs control order along the ranks in which they are stationed. To assist them, Dekadarchs should logically be aligned in the same way.

Left corner of the Chiliarchy, showing the Pentakontarchs and the most likely alignment of the Dekads.

That Dekads form up in one long line, one beside the other, can be deduced from looking at the Menavlatoi unit. This is distinct from the main block of troops, and the wings. It forms one rank, 100 men long, and stands to the rear of each Chiliarchy in the initial stages of battle, until it is required.

Similarly, when the breadth of an ancient Greek phalanx was to be made greater, the classical solution was to halve the depth of all files and put the rear half along side the front half, doubling the breadth. Nikephoros, instead, takes one rear rank of soldiers, marches them off to the left or right end of the phalanx, and by forming them into 3 shorter lines, one behind the other, extends the front 3 ranks of the phalanx by 33 men. These examples illustrate a major break from the organisation of classical armies.

Expanding the front.

Commands and communication passed from the front to the rear in a classical phalanx, but the Byzantine manuals clearly show that, by the 10th century, these were passed along a rank, to the left or the right, rather than from the front to the back. The Byzantine units line up beside their officers5.

A possible explanation for the change is this. Nikephoros' battle line is simply the organisation that the multiple units of a classical phalanx would have used when on the march, one behind the other. When such a phalanx arrived at a battle site, each unit would line up to the left of the one it had followed on the march, taking up much time and requiring more time to reorder the line in its new orientation. Nikephoros simplified the process of drawing the army up for battle by making it fight in the same order it marched. Each member of each unit had only to turn 90 degrees on the spot, to be in position for battle. The only other innovation was to draw this long line into a closed defensive formation, similar to a circle of covered wagons, to protect against flank and rear attacks.

3. How large was a Dekad?

This brings us to the third problem. Each rank, or row, of 100 soldiers is commanded by a Hekatontarch. Each Hekatontarch has two subordinate officers, the Pentekontarchs, who each control one half of the line of 100 men. Reporting to the Pentekontarchs are officers called Dekarchs. Theoretically from their name, dekas meaning ten, these officers would be commanders of units of ten, but the manuals are vague at this point, and we have examples that contradict it. In classical Greek armies, based on the binary system, a Dekad could be a unit of 8, or 16 men. While decimal units seem to be implied by the officer's titles and the round figures found in the text, similar titles and round figures were in use a thousand years before to describe units based on a binary Dekad, and these documents were still available to the Byzantines of 1000 AD.

First of all, assuming that a Dekad does contain ten combatants, do the officers count as part of their unit or are they supernumary to it? Is a Dekad made up of a Dekadarch plus 9 soldiers, or does he command 10 soldiers? My assumption is the former, otherwise there could be 100 soldiers, 10 Dekadarchs, 2 Pentekontarchs and 1 Hekatontarch in each row of infantry, making the line 113 men wide. That this was not so is shown by the method in which the phalanx was widened by taking the rear line and tacking its members onto one end of the existing phalanx. In this manouvre, the rear line moves to one end of the phalanx, divides into 3 shorter lines, and lines up along side the front 3 lines. The manual expressly states that this will result in 2 rows of 33 and 1 row of 34 men. In view of the specific numbers involved, it is unlikely that a Hekatontarchy could be any larger than 100 men.

If Dekadarchs are not supernumary to their commands, there are still a further 3 men to consider, the Hekatontarch and his two Pentekontarchs. If Dekads are units of 10 men, and there are 10 of them, the row of soldiers will be 103, not 100. It is possible that these 3 officers each fulfilled the duties of Dekadarch to a unit of 9 men, as well as their higher officers duties. This would give the 100 man figure of the text, but make them much busier. There is, however, another alternative.

Suppose that each Dekad were to be comprised of 8 men; a Dekadarch and 7 others. Six Dekads would be the command of Pentekontarch, thus making 49 men inclusive, and two Pentekontarchies would be the command of a Hekatontarch, making 99 men including himself. To make up the number 100, I would suggest an assistant to the Hekatontarch.

Greco-Roman armies were traditionally built on the premise of a basic unit of 8 combatants, but there is also minor evidence for this from the Byzantine texts themselves. Digital organisation is confirmed by Ouranos6, where every 2 combatants share a baggage animal, and every 4 combatants (2 groups of 2), share a servant, who looks after their two animals during battle7 . Groups of 2 can form a unit of 10, but groups of 4 cannot.

The method of extending the width of the phalanx also implies a Dekad size of 8 combatants. When one Hekatontarchy is moved to the end of the phalanx it divides neatly into 2 rows of 33 (4 Dekads and a Pentekontarch), and 1 row of 34 (4 Dekads plus a Hekatontarch and his assistant). If Dekads had a size of 10 combatants, it would have been more convenient to divide this lot into 2 rows of 50, or 5 rows of 20!

Support also comes from the earlier Strategikon of Mavrikios (6th century). In his time, infantry was organised in files, 18 men deep consisting of 16 combatants and 2 servants. These infantry files could be divided in two for combat, and half a file of 16 combatants was called a Dekarchy. The manual goes on to say that cavalry Dekarchies were commonly 8 - 10 men in strength. Coincidently, by combining the 2 servants with the Nikephorian Dekad of 8 combatants, we get 10 men.


Dennis, G. Maurice's Strategikon; handbook of Byzantine military strategy. UPP: Philadelphia. 1984.

Dennis, G. Three Byzantine military treatises. Dumbarton Oaks: Washington. 1985.

McGeer, E. Sowing the dragon's teeth: Byzantine warfare in the tenth century. Dumbarton Oaks: Washington. 1995.

Go to Dumbarton Oaks to learn more about Sowing the Dragons Teeth


1 McGeer, 1995, on pages 203-4.

2 Menavlion = a heavy spear or javelin used to fend off cavalry charges. PB.

3 McGeer, 1995, on page 202.

4 = ‘Campaign Organization and Tactics’, Dennis, 1985, pages 241-335.

5 As the chiliarchy (or taxiarchy) itself may have been a relatively new formation, such an innovation in command structure may have been among its distingushing features. The term does not appear in manuals before the middle of the 10th century, such as the Taktika of Leo the Wise (c. 900). McGeer, 1995, page 203. PB.

6 Taktika chapter 57, McGeer, 1995, on page 99.

7 Additional evidence comes from De re militari, chapter 4, the infantry squads posted to watch approaches to the camp were eight men; or in the outer watch posts four men, called a tetradia. Dennis, 1985, pages 265-267. PB.


By Graeme Walker, 1999. All rights reserved by the author.
Webbed by Peter Beatson, 1999.