The Five Sided Cut Method

There are several good ways to check the accuracy of a crosscut fence on a saw; the Five Sided Cut process is one of those methods. This method is commonly used with sliding table saws. The benefit of this method is that it is "error additive", fairly easy to do, and gives the tester a useable measure of accuracy. The process can also be used on standard table saws and other machines.

The Process
1. Select a piece of stock to use for the test, something about16" to 24" square is a good size that will give the test a decent amount of resolution. The stock material should be stable and flat for the best results, ¾" Melamine, plywood, or MDF is a good choice. You'll need to measure some fairly small variations in this stock, the best way to do this is with calipers or some similar tool but it is possible to perform the test by "feel".
2. Set the fence perpendicular to the blade (reading ninety or zero degrees as dictated by your machine).
3. Place the stock against the crosscut fence and cut off about ¼" to ½" of material. Note, it is a good idea to have some reasonable amount material on the "waste" side of the blade. This ensures that the blade is in a stable state and that the test is being conducted under "normal" conditions.
4. Rotate the stock 90 degrees by placing the fresh cut edge against the crosscut fence and make another cut. Repeat this step until you have rotated the stock back to the original position. Note, extra care should be taken to ensure that there is nothing between the stock and the crosscut fence to throw off the cut angle even a little as it will affect the test.
5. Make another cut in this position (this is the fifth cut) and save the off cut.
6. Take THIS off cut and mark one end F for forward (this edge was cut first by the blade), and A for aft in the correct positions.
7. Compare the difference in width between the F and A ends; the difference IS the accuracy. For example if one end is 1/8" thicker than the other then the fence is out of square by an amount of about 1/8" over the sum of the sides that were cut. If a ~24" square test piece was used, the fence is out of square by 1/8" over 96" in length (24 x 4 = 96). The method creates an additive error each time the stock is rotated and cut.

You can use this information to determine how much to move the fence to bring the cut more into square. In the case of a 24" square test piece, the fence would be moved about 0.03" measured at a point on the fence 24" away from the blade (1/8" divided by 4 is about 0.03").
Once the new reference is set, the five sided cut test should be performed again to see how close to square the system is with a new correction. It must be understood that at some point in the alignment process it becomes a waste of time trying to dial out a small error, it is an exercise in futility. This is because of the nature of wood to begin with (which changes shape based on its environment), the basic accuracy of machines and blades, the problems in creating a repeatable setup, and trying to reliably measure such small differences in width.