{Printer Friendly}

John David Lewis

Review of Philip Sabin, Lost Battles: Reconstructing the Great Clashes of the Ancient World. London/New York: Hambledon Continuum, 2007. Pp. xxi, 298. ISBN 978-1-84725-187-9.

Philip Sabin, Professor of Strategic Studies at King's College, London, and co-editor of the two-volume Cambridge History of Greek and Roman Warfare,[1] has combined his interest in ancient warfare with his skills in analytical modeling, and directed them to the "Lost Battles" in antiquity. How do we overcome the limitations of missing, contradictory, or implausible information about some of the most important military engagements of the Greeks and the Romans? These clashes are at once well known to us--by name at least--and stranded so far in the depths of memory and history that even their physical locations often cannot be identified with confidence. Unwilling to admit defeat, Sabin's answer is to apply a method of dynamic modeling "to set each battle much more clearly within the context of the general run of other similar ancient engagements, and thereby to highlight which of the various conflicting interpretations are most in line with what we know from elsewhere" (xiii). The result is an engaging and fresh look at ancient armies, terrain, and commanders. But it is also an invitation and an opportunity to reconstruct the battles on a game board, to "play" them ourselves, and thus to test the assumptions behind their reconstructions as well as our own powers of decision-making. Sabin brings strategic studies, military history, and war-gaming together into a synthesis that bridges the unnatural divides between these fields.

Sabin's comparative methodology aims to create a model, applicable to all ancient battles, produced inductively from an in-depth examination of many battles, which can then be applied back to those same battles through simulations. Implicit premises are, first, that there were similarities between ancient events that allow us to fill in holes in our literary and other sources, and, second, that two-way feedback between generalizations and particulars can help us better understand both. Beyond this, Sabin proceeds with two explicit presuppositions: first, that every battle has an element that can be considered using a "mathematical model" that "provides the framework for troop maneuver and combat resolution" but that is superior to other modeling approaches "because it is expressed more in verbal than in mathematical terms" (xvii).[2] This verbal model avoids the problem of computer simulations that require more precision in the inputs than our ancient sources allow. Because the game is to be played with dice, and because we are supposed to adjust our play as the game progresses, there will be a feedback against the outcomes given by the sources, and a chance to re-configure and re-calculate factors that prevent a proper outcome. The model becomes a way to reconstruct the historical aspects of a battle itself.

The second presupposition concerns free will: "[this] element consists of constant decision inputs by the opposing players, which reflects the essence of war as a battle of wits as much as a blind collision of armed masses" (xvii). Sabin recognizes that an ancient army was not a monolithic structure under the strict control of a commander, but rather a group of human beings with a hierarchy of authority reaching down to the levels of junior officers. The model permits decision inputs at all levels of the game, which will add to its dynamic qualities. Players must make their own judgments about tactical problems, just as ancient commanders actually did. The result is far more sophisticated than assigning numbers and values to troops, then writing a program and letting it run. It allows for a range of variations and uncertainties in the composition of armies, troop sizes and capacities, and allows assumptions--including the command decisions of the players--to be tested against the game itself.  A reader who at this point is not certain whether he is reconstructing an ancient battle or designing a war game has missed the point. The object is to do both, to reconstruct the battle in the form of a game, and then to play it. The game is the reconstruction.

In building the model--and creating the war-game scenarios--Sabin starts with the armies themselves, not with the terrain or the decisions of commanders. He rightly opposes the use of static blocks of troops, and stays on his mission to recreate the dynamic qualities of a real engagement. He looks "to base our subdivision of the armies on the actual tactical organization and articulation of historical forces" (17). He adopts a "grand tactical" perspective--concerned with the overall movement of troops on the battlefield--while beginning his reconstruction of the armies themselves at the unit level. He attempts to create not only a unit-scale by which to gauge the size of the units, but also a scale of values to reflect their quality. Such scales can later be used to compensate for difficulties of terrain or other impediments to good troops. Troops can be understood in three unit classes, each assigned a mathematical level of fighting value. "Veteran" troops = 4; "Average" = 3; and "Levy" = 2. Further subdivisions are between infantry and cavalry, skirmishers, chariots, and elephants, each of which can be modified by a numerical value. One then proceeds by trial and error, considering the evidence of the sources, assigning the troops and values appropriate to the scenario, "adding up the fighting values of the opposing armies and seeing how they compare" (21). Trial and error may seem to be an inappropriate way to begin, but it is perhaps closer to what ancient commanders did than any set program of movement could be.

The armies are projected onto a physical layout, with movements proceeding over time, and with the decisions of the commanders and officers brought to bear. In starting with the army rather than terrain, Sabin assumes "the highly formulaic nature of ancient battles" (29). Here the problems of reducing the complexity of a battle to a game board become manifest, and one must suspend disbelief and go with the flow of the game:

Now that we have come to grips with the unmanageable complexity of real armies and real battlefields by subsuming them within a small number of standardized units, unit types, zones and terrain types, the really hard work has been done, and providing for the movement of troops across the field is actually little more complex than regulating the movement of chess pieces across a board" (34).

Sabin is both overstating the solution his game scenario brings and accurately recognizing its own limits. He reduces the patterns of movement, for instance, to orthogonal directions: "Diagonal movement would complicate things ... and would introduce far more flexibility than is realistic for the unwieldy formations of the time ...." The rightward drift of hoplite armies, and the rightward turns of armies, will not be followed, but will be handled "by arranging initial scenario deployments accordingly" (35). Time is reduced to a series of "turns" (begun with a roll of the dice) and movement to a single "zone" (chariots and cavalry get two zones, except in difficult terrain). 

In dealing with the issues of fighting and command, Sabin completes the battlefield model. It has twenty zones, ten for each of the opposing armies, five zones across and two deep for each army, with forces arrayed within the zones as accurately as can be reconstructed, and a scenario ready for testing against the known outcome. We are now set to play the game, but it has become difficult to see how we are in a position to learn anything about ancient battles beyond the analysis of army units and study of their possible positions on the field. Although the "grid and movement systems" remain "resolutely broad brush," as required by the "grand tactical focus" (41), the mathematical model for troops and maneuvers has reached its fundamental limits. The second element now comes into play: the decisions of commanders and officers. Here in the battle of wits is where Sabin's model might allow us to learn something important.

One of Sabin's assumptions is that "ancient armies were not animated by a single guiding hand, in the same way as chess pieces or playing cards are deployed by a single individual. Instead, the armies were composed entirely of animate entities in the form of individual human beings," who were distinguished from crowds by a hierarchy of officers (61). This left much latitude for local decisions, especially given the vagaries of terrain and limitations of communications. The game allows for tactical input and decision-making at every level. Here this method of conceiving ancient battles bears fruit in active intellectual engagement with a fast-changing collision of forces.

After armies have been assigned to the field under a series of acronyms--"AHI" for Average Heavy Infantry, "LLI" for Levy Light Infantry, "UL" for Uninspired Leader, etc.--play can begin. Cannae is the first battle to which the model is applied, and Sabin works through one possible series of movements. There is a strong reliance on chance here, because the number of moves is determined by a roll of the dice, combined with an initial numerical fighting value. Playing the game places the Romans in a position of inferior generalship, and motivates them to attack first, in order to avoid the horrific envelopment that everyone knows is inevitable. By fighting a battle this way, each side can test the accuracies of the assumptions, the possibilities of movement, and the effectiveness of the decisions. The Cannae game reinforces the importance of the Punic cavalry, demonstrates much more dynamism on the Roman side than a mere envelopment of their static line, and adds a level of desperation to their attempts to avoid defeat. The game brings a certain sense of realism to the battle that is lost in merely reading the sources passively. A fast-paced game may unfold here, as players refer to sources and adjust their armies and their play against their opponents, measuring both against the outcomes.

Sabin provides brief (two- to four-page) descriptions of thirty-four battles, from the Greeks at Marathon (490 BC) to Caesar at Pharsalus (48 BC), along with color maps of deployments, and troop levels and values. As regards, for instance, 1st Mantineia (Spartans and allies vs. Argives, Athenians, and others, 418 BC), Sabin draws on information from other events as he creates the model, in order to address the central problem of reconstruction--the size of the Spartan army. Thucydides' claim, "that the Spartan army was the larger and that there were only a few thousand Spartan hoplites" (105), would require the integration of non-Spartans into the battle line.  This is precisely the kind of issue that the model can test, by varying the appropriate factors and their movements with decisions taken during the game. The result may support Thucydides' calculations (as the Cannae reconstruction showed the plausibility of Polybius's account). Sabin never overstates his case or claims to have solved problems that scholars disagree on; his goal is to stimulate engagement and debate and perhaps suggest new approaches to old problems.

Sabin's energetic approach to the lost battles brings together the researches of an ancient historian and the dynamic thinking of a military strategist in a game of wits. In terms of production, I do wish the book had an index to facilitate battle comparisons. As a classicist and not a strategist, I am intrigued by the possibilities of the model, especially its flexible capacity for feedback and the testing of different assumptions, within its inherent limits. The value of this book is not in any ground-breaking new revelations about these oft-studied events, but rather in its stimulating engagement with the process of managing and directing forces on a battlefield. Many readers, not inclined to pore over the original sources and engage in the debates they engender, may be drawn into the process of military strategizing and execution, and, in the process, discover solutions to the problems that have resisted the efforts of historians. Teachers may find in Sabin's battle modeling an excellent means to engage their students in ancient military history.

Duke University


[1] Cambridge: Cambridge U Pr, 2007, with co-editors Hans van Wees and Michael Whitby.

[2] Cf., e.g., the approaches of Trevor N. Dupuy, Numbers Predictions and War: Using History to Evaluate Combat Factors and Predict the Outcome of Battles (Indianapolis: Bobbs-Merrill, 1979), and Stephen D. Biddle, Military Power: Explaining Victory and Defeat in Modern Battle (Princeton: Princeton U Pr, 2004).