If you go to the office store or hardware store and buy a ruler, you might think that you’ve gotten two for the price of one: a length-measuring tool, and a straightedge. Unfortunately, this isn’t always the case. A regular old ruler may or may not be straight. Try this: put the ruler it down on a sheet of paper, mark a line along the edge with a 0.5 mm mechanical pencil, then flip it over and mark another line on top of the previous one. If the lines are directly on top of each other, the edge is straight — at least, it’s straight enough that the error can’t be seen with half-millimeter pencil line. I have at least one ruler where the gap between the lines is over 1 mm.
Nothing is perfectly straight. The only question is how far from straight your straightedge is. For most drawing and measuring applications, if it passes the test above, it’s good enough. But sometimes you need it to be better. In woodworking, the surfaces of some tools need to be very flat — much flatter than 0.5 mm. A hand plane with a sole that’s flat to only 0.5 mm would be worse than useless; it would unpredictable and inconsistent, and would gouge the workpiece horribly. (Unfortunately, inexpensive metal hand planes are all like this out of the box — maybe not as bad as 0.5mm, but bad enough that they really can only be used to damage wood. Seriously. This is not an exaggeration.) I don’t know exactly how flat the sole of a hand plane needs to be, but it certainly needs to be better than 0.5 mm, or even 0.1 mm.
I was considering buying a good straightedge, but before I actually did it, I saw in Toshio Odate’s book Japanese Woodworking Tools a method for making a straightedge. The purpose of the straightedge in his book is specifically for checking the sole of a plane, but it could be used for any purpose.
To make it: the short version of the story is that you take two pieces of wood, put next to each other and plane them, then “unfold” the two objects so that the planed edges are facing each other, then check for light between them. If there’s light, you shave away the high spots and check again. The are two reasons why you make a pair of objects: first, you don’t need a good reference straightedge to check to see if your new straightedge is actually straight, and second, any deviation will be doubled, making it possible to see errors that are half the size of what you would be able to see even if you had a perfect straightedge to compare it to. This paired-planing method is, in theory, twice as accurate as making a single object and comparing to perfect straightedge.
Planing the two parts of the straightedge. There are two thin pieces of wood, clamped together so they don't fall over.
Checking for straightness. The pieces are "unfolded" so that the planed sides are facing each other. In this photo, the ends here are bowed apart, indicating that I put too much pressure at the end of the planing stroke. I eventually got it much better, but not perfect.
It turns out that this process took more time and effort than I expected. Even with a very good plane, I tend to take off too much material at the beginning and end of the stroke. It took me many tries to get it very straight, but even now there’s a tiny, tiny sliver of light visible in the middle. Still, it’s good enough for my purposes.
The smallest gap I can see between the jaws of a micrometer is just under .001″ (.025mm). If you can’t see any gaps between the two pieces of wood, then each of the two straightedges should be good to half that, or .0005″ (about .012mm). If there are any visible gaps, they can be tested with a feeler gauge. If you don’t have feeler gauges, you can use a sheet of paper instead. A sheet of 20 lb. paper is .004″ thick, so if it won’t fit in any visible gaps the two pieces of wood, then each piece is good within .002″. Keep in mind that the pieces of wood must be held together lightly when checking for gaps — it’s possible to bend the pieces of wood these minute distances if you press them together with any significant amount of force.
This is the smallest size gap I can see through the jaws of this micrometer, at just under .001" (or .025mm). It looks larger in this picture than it does in real life.
One of the pieces of wood is what I’ll call the measuring piece — it’s the one which you use to check objects for straightness. The other one is a companion piece, which is used in the truing process if it ever needs to be done again. A hole is drilled partway through each piece and a snug-fitting dowel is glued in the companion piece. You just push the two pieces together for storage, and pull them apart for use. The piece that’s used for measuring has a chamfered edge so that the contact edge is a little narrower, and so that it’s obvious which side is the straight one.
The two parts of the straightedge. The companion piece has a dowel glued into it; the measuring piece has a hole that fits snugly with the dowel.
The finished straightedge. The measuring piece is chamfered on the straight side. (The other side, facing down, isn't as straight.)
Overall, the straightedge is about 16″ long and each of the two pieces is about 1 1/2″ wide and 3/8″ thick. If I did it again, I’d probably make them a little thicker, like 1/2″ to give them a little more heft and stiffness.
The great thing about this process is that you don’t need a straight surface to make one. You just need your eyes and a decent plane, but not necessarily a great one. With a less-than-stellar plane, it just takes more time and care to remove material from the high spots.
Usually when you want to make sure that you have a straight object, you compare it to a reference object — one which is known to be very straight. The best you can hope for is that your test object is as good as the reference object, but in reality, it’s likely to be a little worse. If you were to make a series of straightedges, basing each one off of the previous one, it would be just like the telephone game: a little is lost every time a copy is made, and the last one would almost certainly not be as straight as the first. What I find interesting about this homemade wooden straightedge is that it serves as the reference object for itself; you don’t need a better (and more expensive) object to compare it to. Additionally, it gives you twice the resolution for detecting deviations from straightness.
The more I learn about woodworking, the more I’m impressed by the cleverness of craftsmen (OK, craftspeople) in the past. There is a huge wealth of knowledge that wasn’t written down because it was passed down from master to apprentice, because some techniques were trade secrets (think of the Italian violin-makers), and because the people who made things for a living generally weren’t the type to write books. In today’s world, the master-apprentice system is a shadow of what it once was, and so this knowledge that is stored in people’s minds needs to be transmitted a different way. Fortunately, we can now learn from books and from other people on the internet.
In the environment I usually inhabit, knowledge and intelligence are highly valued, but only when they relate only to abstract, intangible things. Understanding the physical world, knowing how to make things, is considered quaint and sometimes looked down upon. But to me, the knowledge embodied in this simple object — a wooden stick that is straight and true — deserves a great deal of respect.
Investigations 
Fantastic, just fantastic. The knowledge your average 10 year old had on a farm just 100 years ago could fill books. The very basic skills needed simply to do daily chores has been all but forgotten. The ability to program a satellite for geosynchronous orbit is a feat in itself and deserves its own respect but unplug the world and that 10 year old farm boy would instantly become king of the day.
This is just fantastic, I don’t have expensive tools or even a workshop and I never knew how to get a piece of wood accurately straight. Thank you for taking the time to share this knowledge. Oh and I did grow up on a farm, but a carpenter I am not!
An additional hint: an old “dodge” used for overcoming the tendency to snipe (cut excessively) the ends of wood being planed by a power planer is to simply make the piece a couple inches longer then cut it to the intended length instead of playing around trying to get it perfect end to end. The same approach can be used here if your only problem is trimming down the end(s) too much at the end of the planing stroke: just make the straightedge bigger than you need and cut it off. In production this technique is terribly wasteful, but here you’re building a one-off that presumably will last for some time. Hope this is helpful.