Friday, March 30, 2012

Accuracy Versus Precision

“In theory, there is no difference between theory and practice. In practice, there is.”
-Yogi Berra
I was going through another back issue of Fine Woodworking today, (FWW #25, November/ December 1980) and came across an article titled “On Precision in Joinery,” by Allan J. Boardman. Boardman didn’t specifically mention it, but there is a difference in some fields between accuracy and precision. It’s possible to be accurate without being precise, and it’s possible to be precise, without being accurate. And I think this is a discussion that is relevant to woodworking, too.

In our context, precision is defined in terms of exactness and fit. When joinery is cut precisely, it fits without strength-sapping gaps. Shoulders close up on mortise and tenon joints. Miters come together all the way around. Dovetails fit perfectly. And the assembly comes together, without needing an I-beam clamp’s tectonic level of force to close everything up. (I love I-beam clamps. But they are a bit excessive.) This doesn’t imply that the entire assembly is made without error. Precise work involves making adjustments that take those errors into account, but still allows everything to come together well. Some dimensions may be slightly off, or the wood may have a slight spring in it. But joints that have been precisely fit will still be solid, and the finished product will be strong.

Accuracy is defined in terms of tolerances and deviations. “Accurate to a tolerance of +/- .00004 smidges,” and so on. It’s not about hitting the mark right-on. It’s about how close to the mark you can reliably get, within a margin of error. And that error is the difference between the two words.

The article explained pretty clearly the same philosophy that we learned at North Bennet: Use machines to do bulk waste removal, and hand tools to precisely fit things together. While the method wasn’t specified as such in the article, It’s what’s currently known as hybrid woodworking: Bulk work with machines, fine work with hand tools.

Hand tool purists may not like the dust and mess and noise of power jointers and planers and table saws. But they still work initially to the tolerances allowed by broad axes, froes, scrub planes, and cross-cut saws, when they’re doing the rough dimensioning.

I think that in a hybrid work environment, it’s fair game to use whatever method you want for bulk waste removal, including combinations of hand and power tool methods. I’ve had gnarly figured planks that I didn’t want to run across a jointer, for fear of massive tearout. So I’ll flatten one face with a jack plane that has a cambered iron, accurate to a margin of error that’s defined by the scoops I took out with the plane. The high points will all be flat enough, and with my roughly jointed face down, it goes through the planer just fine. It’s accurate work, done to coarse tolerances. And the thickness planer doesn’t know or care. Once the other side has been surfaced, the board gets flipped, and my plane strokes will be removed during the next pass.

But however you hog off bulk waste, both camps can agree, I think, that there’s a time to move from accurately used rough work, to doing precise work with the final fitting tools. There’s a convergence in method here, and however you choose to work, the transition between accurate and precise methods is where the magic happens. It’s worth paying attention to, because so much time can be lost.

Precision should take over when the limits of accuracy have been reached... but not before. If you have a fetish for hand planes, and you find yourself taking off an eighth of an inch, 2 thousandths of an inch at a time, you’re going to be at it for a while, and the odds that human error will be introduced will increase. That’s a lot of wasted time, with the potential for wasted time and effort thanks to a part that's ruined after chasing a final dimension around for too long with a hand plane.

In a similar vein, accuracy should be taken to a reasonable degree, but not past that. While high levels of accuracy are very helpful, it’s not always appropriate to try to substitute high levels of accuracy for precision. Margins for error are measured in reference to an ideal, and not necessarily to the conditions present in the project at hand. Highly accurate jigs and fixtures will not necessarily bridge the gap from being accurate tools to becoming precision instruments, because adjusting will still be needed on the back end to compensate for errors in the project.

Some woodworkers will try to reduce the margin for error to zero on their jigs and tools, in ways that aren’t always appropriate, especially when they’re more comfortable with refining their accuracy than they are with being precise. Shooting boards are a prime example. Some people build shooting boards to excessive levels of accuracy, spending hours with dial indicators and feeler gauges. They’ll insist that it’s as accurate as humanly possible, and that no adjustment should be necessary at all. But there’s a discrepancy there between what’s ideally called for, and what’s really needed. To precisely miter two pieces together, especially when you’re working at a large enough scale, it’s very likely that the final angles won’t be exactly 45 degrees. And if all the shooting board will cut is a perfect 45 degree angle, you may need to shim the fence a little bit here and there with a piece of tape, or veneer, to make precise adjustments to the fit of the joint.

The accuracy geeks think that this is heresy. They believe that their shooting boards shouldn’t need any adjustment at all. Some of them are actually insulted by the notion. Instead, they will insist that all materials must be planed and dimensioned so perfectly that no error will be introduced into their attempt to cut a geometrically perfect joint with their geometrically perfect tool. If there is a problem, they’ll say that it is clearly with the materials, or the project at hand.

I’ve also talked to finish carpenters here in Boston who are so used to working on old buildings that they leave their highly accurate sliding compound miter saws at home. After over 100 years, none of these buildings are straight, square, or plumb anymore. So a saw that cuts a perfect 90 or 45 degree angle really doesn’t help, because the walls aren’t square to each other, or to the ceiling, and the walls themselves may not be straight. The amount of error presented here is huge, and it may be due to the plaster, the framing, or even the foundations. So instead of bringing in a back-breaking sliding power miter saw, they’ll bring in an old hand-saw miter box. And they’ll make adjustments until the miter in the crown molding comes together precisely, in the context of the room that they’re finishing.

Good woodworking is the combined process of accurately making parts, (however you choose to rough them out) and precisely fitting them together. It’s important to develop your ability to be accurate, to minimize the amount of precision work that you have to do. And it’s important to develop your ability to be precise, because even high levels of accuracy will still encounter errors in the real world.

Second quote from the article:

“Precision can, on the other hand, become an obsession that goes beyond common sense to the point of inconsistency with the nature of the wood itself.”

Wood moves. It’s why we use the joinery that we do. It’s why it’s important to look for clear, straight grain, to let wood acclimate to the shop for a few weeks, and mill to final thickness in stages. Over the years, and from season to season, the wood will continue to expand and contract. The traditional joinery that we all know and love was designed to work with the material, and not against it. It’s not ideal, nor is it perfect. But it works.

And that’s a good thing. As woodworkers, we can’t worry too much about an ideal world. The most we can hope and strive for is to be precise in our work, because the best our material will ever be is accurate... give or take a margin of error.

No comments: