The size of Xerxes' army continues to mystify lovers of this historical period and remains an ongoing controversy. Debates on the question of whether Herodotus can be trusted on numbers, spark some of the longest threads in ancient history forums, and vigorous debate in scholarly books, essays, and dissertations. Below I summarize a few key points and competing views.

Generally, manpower numbers from ancient historians enjoy little confidence and invite correction, because most often the historian does not disclose how he arrives at his numbers, which prompts the question of how would he possibly know short of counting the soldiers himself. Generals would have known the numbers of their own army, but the numbers of the enemy must have been left to experienced eyes to estimate. If the historian had access to these primary sources, numbers may be good ballpark figures, otherwise anything goes, and we can rarely be sure of which case obtains.

We stand irremediably at the mercy of ancient sources replete with propaganda and gossip from writers with less than rigorous methods. Therefore, the modern historian feels that failing to give accurate, detailed manpower numbers of armies and battle reconstructions amounts to shirking his duty, and often fills in the blanks or replaces the numbers given by the ancient sources, using a method described as informed speculation. As we will see, this is no small task.

For Xerxes' army, Herodotus gives 1,700,000 infantry, 80,000 cavalry, and 20,000 Arab camel-riders and Libyan charioteers. He adds the supposition that attendants and camp followers must have numbered just as many, but wisely he does not give a grand total. A commemorative inscription attributed to the poet Simonides of Ceos (c. 556–468 BC) gives 3,000,000 as the size of Xerxes' host. It is plausible that only the elite troops and the royal entourage had a large number of attendants, since every relocation of the king and court must have entailed an immense staff, and that there were considerably fewer attendants for the rest of the soldiery, which together with the commissariat personnel, could bring the totals for Herodotus and Simonides into better agreement than the disparate numbers of modern historians.

British Major General Sir Frederick Maurice wrote an influencial paper on the size of Xerxes' army (Maurice, 1930), which later historians have used as a starting point to form their own theories. About a century ago, Maurice was also struck by the peculiar disparity of these modern numbers and not much has changed. I quote: (my bold)

Almost all are agreed that Herodotus' figure of 2,100,000, exclusive of followers, for the army (Book VII. 184-85) is impossible. Grote, while confessing himself to be unable to arrive at any definite figure, considers the army to have been the greatest assembled at any epoch in history; Rawlinson estimates the armed force at 1,190,000; Thirwall is disposed to accept Herodotus' figures; Curtius puts the strength of the army at 880,000; Bury at 300,000; Busolt accepts 300,000, including followers; Grundy accepts half a million; Macan computes the number of combatants at 360,000; while Delbrück, who probably consulted some of his friends of the German General Staff and learned from them the nature of the problem of marching a large army through such country, puts the number of Xerxes' combatants at from 65,000 to 75,000. Recently Mr. J. A. R. Munro has, first of British historians, examined this question of the size of Xerxes' army from the point of view of the military requirements. He has confined himself to conditions of organisation and command, and comes to the conclusion that the Persian army was composed of three corps, each of about 60,000 men [180,000]. I had in 1922 independently arrived at a very similar [210,000] figure by a different process. (Maurice, 1930, p. 211).

Maurice clearly made a reasonable effort to give a sensible fresh number. He even retraced the segment of Xerxes' march he considered most difficult in person, book in hand. See Maurice Findings for my take on the problems of Maurice's conclusions.

This is a less than comprehensive list of post-Maurice historians and their numbers for the size of Xerxes' army, which suffices to illustrate the continuing lack of consensus: Bradford (1993) follows Maurice 210,000; Barkworth 60,000 infantry (Barkworth, 1994) ; T. Cuyler Young, Jr., doesn't give a number but thinks Maurice gave only a maximum possible and the true number only modestly outnumbered the Greek forces (Young, 1980).

There you have the manifest reliability of a few variants of the informed speculation method, which is what one would expect from compounded guess upon guess.

There are several problems with the view that the numbers Herodotus, Aeschylus and Simonides give were a mere act of Greek propaganda: Their motive was to exaggerate the glory of the Greek warriors who defeated them. (Briant, 2002, p. 527). For a cogent rebuttal see Scott, 1915. The proud Persians seem to have neither lodged complaint nor provided correction at no time of this insulting alledged exaggeration mocking their military prowess, in the remaining 150 years of their empire.

A new fact concerning the Persian Wars is that the immense majority of modern writers agreed that Xerxes' army was not really nearly so big as Herodotus represents it. Unfortunately modern authors often spoil a valuable piece of criticism or research by letting professional etiquette compel them to tack on original views of their own with regard to the size of Xerxes' army and these are merely futile, if only judging by results.

The aim of Xerxes-The-Recount Project is most definitely not to validate Herodotus' numbers, but only to show that his numbers were not impossible, and present factors in favor and against their plausibility. Herodotus, as a contemporary source, was well placed to know best, but even as active researcher nearest in time to the events he describes, the question arises of whether his sources regarding numbers were much better than ours.

N. Whatley has this to say:

We lack, then, in the reconstruction of ancient battles, those sources of information which are our chief assistance in dealing with modern military history—written orders, states, diaries and the like. Equally important is the fact that the ancient historians on whose narratives we have to rely were to all intents and purposes as much without this form of evidence as we are. Herodotus, certainly, had practically nothing of the kind; Thucydides very little; Roman historians rather more, because the Romans had more national instinct for preserving this kind of record. (Whatley, 1964).

The very point about Herodotus would depend on what he could glean from Persian sources he most likely consulted, which in highly bureaucratic Persia may well have include the kind of records Whatley describes.

Given the candor of modern historians, who want to cancel Herodotus' numbers, admitting that coming up with a confident new number for Xerxes' host is problematic, it is remarkable how many go ahead and do it anyway. The rigor scholarly work demands ought to compel them to refrain, but the habit of revising ancient numbers or filling the blanks appears simply irresistible, and thus remain determined to provide their best manpower estimates despite a dearth of hard data, revealing a longing for powers they do not possess.

I find little value in this policy because a number with a likely gross margin of error and/or extrapolated from too few data points of non-smooth data hardly qualifies as information. An estimate, after all, refers to a thought process backed up by data, aided by the ability to find out at some point how accurate the estimate was, and the opportunity to apply corrections to the process, leading to a gradual improvement of accuracy. An estimate is akin to a bet, which has two parts: (1) the guessing part of the wager, choosing a number, horse, team, or the cost and level of effort of some task, or some other wager, based on knowledged and experience, if any; and (2) the result, the outcome of the bet, which serves as gauge of the accuracy of the forecast, and provides an example to follow or avoid to learn from. In ancient history, we have only the first part of the bet, but not the second–and are left wanting for a result to judge by. We would have little desire to make bets, and a need for a gambling industry, if we never got to know the outcome of our bets, yet this is the sort of pointless bet historians often make on manpower numbers, which they can rarely, if ever, verify. I disagree with the usual response to this argument, that such numbers are better than nothing, as though we had an obligation to fill mandatory fields in a form, with guesses.

Ironically, Herodotus constitutes the only instance I know where an ancient historian gives both the size of an army and full details of how it was arrived at (Xerxes' infantry census at Doriscus, Hdt. VII 60), together with other consistencies, yet it is he who has generated the greatest disbelief and controversy.

THE PROBLEM OF SELECTIVE EVIDENCE

From the fallacy of the same name, this problem refers to choosing the hits but ignoring the misses, also known as cherry picking. The best historians do it, more or less openly, and Herodotus gets again the full brunt of the vice. When a piece of evidence lies in the way of rejecting Herodotus' numbers, it is simply set aside, while others that would support the rejection are gleefully emphasized. Examples abound.

An egregious case comes from a most admired historian, Pierre Briant, who flat out states that to find a solution to revising Herodotus' numbers: ...that for reasons already given (chap. 5/5), the review at Doriscus should be set aside. (Briant, 2002, p. 527). The Doriscus Census is a most extraordinary piece of evidence supporting Herodotus' numbers, one of a kind in antiquity, explaining how Xerxes counted the units in his army at the Persian base of Doriscus in Thrace. If the purpose was to evaluate the evidence fairly, you would give this crucial piece every chance to live up to the best of its possibilities, instead, Briant discards it for no good reason I can apprehend. Let us examine the reasons presented to justify such dismissal. I expected a fatal flaw I missed in Herodotus' narrative or a cogent train of reasoning the would render the event impossible, but no, no such thing. Briant highlights the extraordinary diversity of the troops from 47 national contingents and their manifold armaments, and questions how could Persian commanders exercise control of such motley crew, to which there are obvious answers. Then, he correctly states that many of these units never saw battle, for which he provides an explanation himself, so that cannot be it. Next, because Xerxes after the census organizes a parade of his troop assortment, he deduces that Xerxes was not interested in his military forces but only in showing the diversity and might of his empire, as if both notions were mutually exclusive, when they are not. I have gone through Chapter 5/5 with a fine-tooth comb and that is how best Briant justifies disregarding an essential piece of evidence supporting Herodotus' numbers. (Briant, 2002, pp. 195-8.)

Herodotus not only gives the method of counting Xerxes' infantry (Hdt. at Doriscus in lots of 10,000 and the total count (170 x 10,000 = 1,700,000), but also gives an Army List (Hdt. VII 60-88), a long list detailing the manner of dress and weapons of each national group that composed the army, which suggests that Herodotus had access to Persian sources that had recorded the participant ethnic groups from all over the empire. Enter Peter R. Barkworth, and his dogged attack on the veracity of this Army List, which he attributes wholly to Herodotus' desire to invent, due to various indications. For example: that the Army List does not match his previous Satrapal List (Hdt. III 90-97). Fair enough, though, if Herodotus desired to invent either list, it would have been easy to made the two list match to conceal his alledged fraud. Another given reason is: Geographically, the Arminians were not contiguous with the Phrygians and thus it seems odd that they were brigaded together. (Barkworth, 1994, p. 157). These and similarly thin arguments serve this historian for the purpose of cancelling the Doriscus Census number, and subtitute his own of 60,000. Of course it is evident that to command a mere 60,000 men, Xerxes would not need 36 generals and six marshals, five from the royal house, or build costly bridges over the Hellespont, so the Army List had to go as well.

We have a wealth of such examples to indicate that the rejection of Herodotus' numbers has become a historical mannerism that leads authors to reach conclusions not by way of true premises from the evidence, but instead they tailor the premises to yield a foregone conclusion. That is, in this case informed speculation is indistinguishable from pretzel logic.

A modern detractor of Herodotus accuses him of selling us a Great Event, which explains the emphasis on King Xerxes and his great powers, the inflated numbers, the rivers drank dry, and the epic accounts. One may ask what would constitute an great even for Mr. Young, if the very king of the greatest empire on earth is coming with a big army to burn your capital and enslave the people, was no great event. Anyway, Herodutos would not be criticizing Xerxes arrogant pride for building the Athos canal, if he was intent on selling anything of the sort: As far as I can judge by conjecture, Xerxes gave the command for this digging out of pride, wishing to display his power and leave a memorial; with no trouble they could have drawn their ships across the isthmus, yet he ordered them to dig a canal from sea to sea, wide enough to float two triremes rowed abreast (Hdt. VII.24).

Though ultimately victorious, the Persian wars were a tragic event that shook the Greeks to their core, but to Persia it was little more than a costly misadventure abroad. If Herodotus and his sources managed to cleverly weave a web of deception in the popular mind and for posterity, and the Doriscus Census, the Army List, the Hellespont pontoon bridges, and other parts of the narrative that support Herodotus' numbers were mere fabrications that need to be set aside, it is remarkable how this conspiracy of lies remained unchallenged by its main victim, Persia, and how it garnered the tacit complicity of the great historians, philosophers and truth seekers of the following decades and centuries; most importantly Thucydides, who was keen to correct Herodotus on minor details; Plato, Aristotle, etc. Their vociferous silence is indeed remarkable. One may wonder whether in this instance silence betokens assent.

There is no reader input in the pages of this booksite. Direct comments to soc.history.ancient Google newsgroup. Start a thread with the word Xerxes in the title. The forum is not moderated.