There are many books which discuss easy home methods for checking alignment. If you're at all interested in this sort of thing, you should get a copy of Fred Puhn's "How to Make Your Car Handle". It's a great book and covers a lot of useful techniques as well as theory of suspension design.
1. Drive the car to a level parking place. Ideally you want the car loaded as it will be when it is driven (e.g. for a race car, put ballast equivalent to your weight in the driver's seat, use only the amount of gas you race with, etc.). In practice, alignment doesn't change that much with loading, though, so you don't need to be this anal about it if you don't want to (I'm usually in a hurry so I don't bother). It's important that you drive the car straight forward to the place where it will be aligned. Once you turn the wheel, or jack the car up and then set it down again, you've changed the static alignment and the numbers you measure will be wrong.
2. (This part I got from Brad Martinson, and it's a really neat hack.) Get two straight pins and stick one into each front (or rear, if you're checking rear alignment) tire on the rear of the tire at some constant height H above the ground. H should be lower than the ground clearance of your car in the neighborhood of the wheels.
2(b). Jean-Yves Meunier came up with an improvement on Brad's method,
which allows you to skip step 3:
In the tape measure I use, I piarced a hole slightly bigger than the
pin heads at about the 6 in. mark (in the middle of the curvature of
the tape).
You can now hang the hole on the tape on the pin head, this allows you
to measure the distance without a helper (less nagging if the helper is
your wife, more beer for you if it's your neigbour)
3. Have a friend hold one end of a tape measure against the tire starting at one pin. You stand on the other side of the car, pull the tape measure so that there's no slack and measure to the other pin.
4. Roll the car forward (not back, you'll roll over the pins!) until the pins now stick out the front end of the tires and are again at height H.
5. With friend, again measure between the pins.
6. The difference between these measurements is your toe.
Notes: you don't need pins (in fact, I seldom use them) as long as you can pick a repeatable place on the tread from which to measure (easy with A008's, difficult with Comp HR's, depends on tread pattern). An easy way to make sure the height H is constant is to use a piece of 2x4 as a marker (4" seems to be a good height on a lowered X1/9; on most cars you could probably measure somewhat higher). Measure as high as you can, because you get better accuracy that way (the tire sticks out more higher up, so you're measuring bigger numbers and your error is a smaller percentage).
Another note: after reading this page, Gerard pointed out that if you roll the car around during the process of checking or changing toe, you should end up rolling it forward, not backward, because if there's any slop in the suspension linkages, you want the toe optimized for the way the suspension slops going forward.
In practice, 1/32" is plenty of accuracy for measuring toe (you don't really know precisely what your front toe should measure to better than 1/32", do you? I sure don't, and I've been experimenting with my car's handling for quite a while. Optimal settings change according to tire compound, my mood, wear in the suspension bushings, how cold and wet the weather is, etc.).
If you want a bit more accuracy, then put the car on alignment plates. Get four thin squares (about 6"x6"x3/16", say) of smooth metal. Use them to make two sandwiches of metal outside, oil or grease inside. Now put one sandwich under each tire. If your metal plates are thick, then put something of equal height under the other end of the car, so it still sits level. Now you have an alignment plate just like the professionals use (only a lot cheaper) and you can turn the wheel, move the car, etc. without screwing up your static toe measurement. You can't use pins or roll the car back any more, though. [ George Tylinski suggests using waxed paper, which also avoids the risk of oil or grease contacting the tires. I haven't tried this yet, but it sounds like a fun idea. ]
A method I like better is to drop a vertical (using a plumb-bob) from the centerline of the front and rear of each tire. Do all four tires while you're at it. Even if the rear is non-adjustable, you might as well check it, because this will tell you whether your frame is straight, so it's a good thing to do on a newly-bought car. Make a mark on the floor of the garage below each tire centerline. Now drive the car away and draw lines all over the place and measure all of them: you can get front track, rear track, right and left wheelbase, front and rear toe, and diagonals between the tire centerpoints. Comparing things like the diagonals and the right/left track will tell you how straight your frame is.
If you're measuring camber with the intention of changing it if it's wrong, do that BEFORE you measure toe. On most cars, changing camber will also change toe (but not vice-versa), because the angle of the struts or other suspension pieces which affect camber is not exactly vertical.
Get a carpenter's level. You want one which is slightly shorter than your wheel size (e.g. on my 13" rims a 12" level works okay). Rest the bottom of the level against some repeatable part of the wheel (like the bottom of the outside edge of the rim, but you'll need different techniques with some wheels). Now swing the level (without losing contact with the wheel) until it's exactly vertical. Using a caliper (you can use a ruler, but a caliper is much easier) measure the distance from the level to the wheel. Now you have your camber in inches. Some trigonometry can give you the camber in degrees, if that's what you want.
There are lots of variations on this. You can use a coarse-thread screw instead of a caliper, and count turns. You can use a dial indicator instead of a caliper. You can make a gizmo with sliding fingers that stick out from the level and contact the wheel at just the right points for your size wheel. You can just hold the level against the wheel and make calibration marks on the window where the bubble appears. You can use a vertical from the floor and measure from there to the bottom and top of the wheels, then subtract. Whatever seems easiest for you.
Negative camber can help the tires from tucking under (SCCA requires that cars with swing axles, like Spitfires, have significant negative camber to prevent the wheels from tucking under and perhaps coming off the rim and causing the car to roll). But the main reason most race cars have negative camber is to compensate for decambering under load.
What you really want is to have zero camber at all times. When the camber is zero, the entire portion of the tire's tread is in contact with the ground, and the more rubber you have on the ground, the more traction you have. Sounds simple, right? But when you add a cornering load, you get more positive camber on the outside wheels (imagine the whole car tilting so that the inside wheels come off the ground. Notice what's happening to the outside wheels now.), and with most suspensions (e.g. MacPherson struts) this effect is made even worse because body roll affects the suspension geometry to make the outside camber even worse. That's why race cars tend to be very stiff. You want to minimize body roll to minimize the decambering effect.
If you set the car up with lots of static (i.e. with the car sitting still, no cornering load) negative camber, then when you get into a hard corner, the suspension decambers and voila!, you have zero camber on the outside wheels, which means you maximize your cornering traction. Only one problem here: when you're going straight, e.g. when you're braking in a straight line, you have no lateral force on the car, and since you've set up the car for so much static negative camber, only the insides edges of the tires are touching the ground, so you get poor braking performance. This also applies to acceleration, if you have a car with enough power that traction is a problem on acceleration. With my X1/9, I wouldn't know about that. :-)
When you set up a car, you have to make a trade-off between straight-line traction and cornering traction. If you make the car super-stiff, to reduce body roll, then you need relatively less negative camber so you don't have to compromise as much, but then you have trouble if you want traction on bumpy roads. Using swaybars, you can get more roll protection with less stiffening, so you're better on bumpy roads, but then you've compromised the independence of the suspension and you may end up with too much wheelspin accelerating out of turns. You also have to take into account the tires: tires with soft sidewalls tend to roll under during cornering, so they need more negative camber. Bias-ply tires, like Hoosiers, with their stiff sidewalls, work with far less camber than radials. On the X1/9, Hoosiers want less than 1 degree static negative camber, whereas A008R radials want about 3.5 degrees! You can tell by looking at the wear patterns. Hoosiers will wear first on the inside edges, A008R's always wear out on the outside first.
Setting up any race car is a trade-off and you have to decide what factors are most important to you. My personal choice is to use radials (I don't like the way Hoosiers lurch), as much camber as I can get (currently about 3.5 degrees until I get the time to fabricate camber plates), super-stiff springs (I don't mind that the car gets airborne over bumps. In fact, I find it fun !) and no swaybars (I had major wheelspin problems with a rear swaybar, and none at all without it. Besides, swaybars add weight.). I leave the caster alone (never even measured it), use front toe to adjust turn-in characteristics and use rear toe to adjust oversteer/understeer balance.