This historical simulation from the author of Napoleon At War is set in the American Civil War, and recreates the practical difficulties of the battlefield. Using much the same system as Napoleon At War, Yankee focuses on two battles, loaded as separate programs - Gettysburg and Chickamauga.

The American Civil War is probably best known through Gone With The Wind and other popular romantic portrayals: as a war, fought over issues and principle. It's probably not alive in the British imagination. This detachment means a computer warp me based on the war could easily be reduced to an exercise without atmosphere - particularly when, by presenting small-scale battle simulations, the game gives no sense of the overall shape of the war.

Yankee avoids this by taking a literal historical approach and backing itself up in the rules with detailed descriptions of the battles (I would have liked to see some general background as well).

The Battle Of Gettysburg took place over four days in July 1863 and Chickamauga over two days in September. Unfortunately there's no two-player option, and the gamer must take the Union side in Gettysburg and the Confederate side in Chickamauga. (The program's artificial-intelligence routine extends to the player's units too, so I can't see why there's no choice of sides.)

Landed with being General George Meade in Gettysburg, the player has command of six corps. The corps, of three divisions each, are under the command of generals with names like Sickle and Sedgwick. Corps command is central to Yankee, as it was to Napoleon At War. You can give separate orders to each division, but the game really works through the command unit; the other two units in the corps mimic its movement orders and throw themselves untidily into the fray.

But it's difficult to predict where all three units are going to end up if one of the divisions gets bogged down in difficult terrain or collides with another unit. And when there are two or three corps trying to attack a block of enemy units they're unlikely to arrange themselves sensibly. This can be very frustrating - but war really was frustrating.

The map isn't large - about three times the size of the TV display, the author claims - but it's clear and detailed, and fills the screen without the clutter of extraneous information. There are eight types of terrain, making up an interesting landscape which seems to be to the correct scale. It's easy to relate the size of the square 'counters' to the terrain features, which are helpfully illustrated in the rules.

Terrain is important to the gameplay, and most features of the landscape give an advantage to the defender (in Gettysburg the best place to defend is a wooded hill). This increases the variety and sophistication of tactics enhances the realism and so helps generate that important and indefinable quality, atmosphere.

There are only three types of troops available: infantry, cavalry and artillery. The cavalry can move faster than the infantry units, but don't seem to have a dramatic advantage in combat. This lack of variety is doubtless historically justified.

Artillery units are not under the command of any corps leader and have to be moved independently; they're difficult to manoeuvre and fire, but the computer opponent is disturbingly adept at using them. They're also vulnerable to enemy attack and can be wiped out in a few turns.

The game operates on a simple two-stage turn structure. You give orders at leisure via a series of option menus, and there's then an action phase in which the orders are executed and combat is resolved. Combat occurs automatically between adjacent enemy units, and as In Napoleon At War strength is (crudely) knocked off 500 men at a time. It's possible to work out how many strength points any unit will take into combat, from their numbers, which can be checked in the orders phase, and their morale, which gives bonuses in seven stages of cheerfulness from Excellent to Abysmal.

But you can't be certain which square or terrain your units will end up fighting from, or whether they'll be defined by the computer as attacking or defending. This element of uncertainty is nicely judged, between chaotic randomness and boring certainty. The combat is shown unit by unit, and it lacks in impact - 500 men go to their deaths in unspectacular silence.

Before the action phase starts, the player may get messages from the corps commanders in a touch of atmosphere-building (again, as in Napoleon At War). But soon the generals, whining for advice, become irritating and one's tempted to send them in against impossible odds just because they keep complaining about any odds at all!

Enemy units are hidden from view till they come within reconnaissance distance of the player's units. I find hidden movement disconcerting and unreal, and Yankee made me realise why: when the enemy side moves, the program doesn't show the 'out of sight' units but still scrolls to where they are on the map - so you watch the map shifting meaningfully about in blankness.

True, a real general wouldn't know where the enemy was, but he wouldn't have an omniscient view of the surrounding landscape either. The programming gaffe in Yankee underlines this.

But on the whole hidden movement is used effectively here; the reconnaissance scope of the units is wide enough to make sure you don't end up planning attacks against invisible regiments.

The rules are of a high standard, of content if not of presentation; the instructions explain them well, the two battles are described and set in context: and the designer gives an account of himself - which might make you more sympathetic to the game's Idiosyncrasies!

Yankee has a deceptive complexity beneath a simple command structure and onscreen appearance. If you have sufficient self-control to restrict yourself to giving corps commands the game is very difficult - I was wiped out twice on the easiest level - but there is hope of improvement with practice.

The play is perfectly paced, and the two scenarios are very different - Gettysburg is defensive and Chickamauga is offensive, for a start. Though you can't choose sides, which is a limitation, Yankee should have lastability. It's certainly one of the best battlescale simulations I've seen, and all wargamers - particularly those who like a traditional approach - should enjoy it.