Last week I received another of the never-ending stream of forms customers keep sending to update their records. I filled in all the boxes, of course; if you don't, some clerk will be all over you because they don't understand what the words "not applicable" mean.

They included two pages of fine print material that reminded me to get new glasses. This was their purchasing terms and conditions that would apply on all future orders.

If you took all of this legal mumbo jumbo seriously you'd never process an order from them. Because lawyers create this legalese, some time ago I asked my lawyer whose terms count in such matters. A few paragraphs later he basically said it depends on a judge if a fight ensues.

What caught my eye was the customer telling me what my warranty must be-one year for everything including service. Because we provide calibration services for this particular company, I wondered if they really meant it. In thinking about it, I realized that while many companies don't go to such lengths, they do think that calibration results are good until the next time, which may be a year or two down the road.

The bells go off when the numbers for this year's calibration of an item are different from last year. Some folks consider such differences an indication that the calibration was not done correctly and ask that the work be done again under warranty. This possibility always exists but there are other reasons that may account for such differences.

One reason the numbers change is because the gage or instrument involved has changed. I realize that's a pretty off-the-wall statement to make because no one uses his equipment enough for that to happen. Just ask them.

Many things can happen to an item to change its size or performance. In the case of a plug gage, for example, if it's made of steel, it can grow or shrink on its own without outside influences. Most of the time such changes are barely discernable in use but can appear as changing data on calibration reports.

Another reason the numbers can vary is measurement uncertainty-probably the most common situation. Unfortunately, many people do not understand this and assume the laboratory is trying to baffle them with the subject to avoid redoing the work for free.

If the lab does the work again and the next set of data more closely matches last year's numbers, the customer feels vindicated in their claim that the work was not done correctly. If a repeat of the calibration comes up with data that closely matches this year's numbers, the customer will complain that the work of a year ago must have been poorly done. I believe this is what is called a no-win situation.

Many customers think they're paying big bucks for numbers that have no errors attached to them-

an incorrect assumption. Trying to convince them that no laboratory is free of the measurement uncertainty situation is not easy. This is especially so when some lab that doesn't know about uncertainty reports to the customer its repeatability as its measurement uncertainty. A small repeatability number beats a higher uncertainty number every time.

I once tried to drag a Roman, Pliny the Elder (70 AD), into such a discussion with his observation that "The only certainty is that nothing is certain." The customer was not impressed.

One cause of the problem is lack of experience with dimensional measurement. The numbers don't mean anything to many people. But if you've been doing dimensional measurements yourself as a tool maker or machinist or an inspector, you gain an appreciation for the fact that a wide range of factors can wreak havoc on the best of measurement situations.

Basically speaking, calibration results are data that was valid at a specific point in time. Yes, I'm aware that time is an illusion, but I won't go into a lengthy discussion on that. I merely use it as a convenient form of reference. You can check with your local physicist on that subject.

After the calibrated item leaves the lab, all bets are off and your warranty has expired.