A lot of bakers and chefs and chemists will tell you that they’re in the same business. So, in a way, are bartenders and cheese makers and drug dealers and brewers and vintners. Because we all need to measure things a lot and we need both accuracy (the ability to measure amounts reliably) and precision (the ability to consistently measure the same amounts).
When you get right down to it, precise, accurate measurement is terribly difficult to carry out, and it’s critical to turning out a consistent product or sale every single time.
In undergraduate chemistry, many students learn a discipline called “quantitative analysis” and it treats an issue that most scientists, including many experimentalists, gloss over as they pursue their science. It’s called error propagation. It’s the term for mathematically propagating the known inaccuracies of any measuring instrument (and each instrument has inaccuracies and imprecisions) through the total experimental calculation so you know to what degree to be confident of your answer.
Another part of quantitative analysis is the discipline of using statistics properly, and of doing experimental design well. A good experimenter is confident of her results. Her peers can know precisely how confident she is and in turn how reliable those results are.
All of this boils down to wanting to know that our measurements are consistent and useful and that they tell us as much about the truth that we’re measuring as possible. Additionally, quantitative analytical techniques give us a strong sign of how much we can truly rely on the quantities we measure.
These same concerns affect the experimentation, research, and products of chefs, cheese makers, brewers, vintners and others involved in precise and accurate weighing and measuring. Depending on the field or discipline, they matter to a greater or lesser degree of precision. This variance just indicates how fine the measurements must be to support the work being done. Scientists and bakers and some specialists in every field need VERY precise measurements. In some disciplines where art is as important or moreso than precision and accuracy, there is more leeway.
First, let’s have a look at a good TED-Ed treatment on accuracy and precision:
For those with no time for a parable about William Tell here’s the summary:
Non-technical folks often use the terms “accuracy” and “precision” interchangeably. But actually, these terms are about two different but related factors in measurement or in performance.
- Accuracy is how close to the real value the average of a set of measurements will get. The cluster might be spread quite widely, but the overall average is correct to the quantity actually measured.
- Precision is how tight a cluster a set of measurements have. The cluster might not be close to the real value but it predictably misses by a predictable amount and direction (or vector) every time.
In most cases, scientists and other users of measuring instruments want to have both high accuracy and high precision, which is why, of course, less technical folks use the terms interchangeably.
But in practice we often have to settle for less. Try buying a kitchen scale while pushing the boundaries of what’s considered normal accuracy/precision. For reference, precision better than 0.05 oz or 1.5 grams is unlikely in a kitchen scale that has a total capacity of higher than a kilogram or 2.2 pounds.
What you’ll get for reasonable amounts of money is scales that can measure down to 0.05oz with precision enough to usually weigh within the same 0.05 oz if you repeatedly pick up and set down an item. To get any better, you usually have to measure much smaller things on scales built for much smaller quantities and/or spend crazy amounts of money.
One issue relevant to wanting high precision and consistency is that you have to train your technicians (or, in a baker’s case, he himself or, in your case, you yourself) to consistently take measurements in exactly the same way each time. This tends to contribute to improving precision. For example, if you learn to measure out a cup of flour by always sifting into a volume measuring cup and then leveling it off by scraping the top lightly with a straight edge against the top of the cup measure, you will have a relatively high precision across different measured cups. If however you eyeball the level or don’t always sift into the cup or don’t always use a straight edge, then your precision will suffer over the many times you measure out flour.
The video mentions a way to fix accuracy problems. It’s a formal process called calibration. If you know any gun owners you may have seen or heard of them calibrating their scopes. Returning to the idea of using and calibrating a scale, most manufacturers of kitchen scales don’t bother with this, but it is very important for scales that measure smaller amounts. It’s also a great idea to calibrate your oven, if you have time, and maybe your candy thermometers. All of these calibration efforts require a trusted reference measuring device.
With scales, calibration is relatively easy. You can buy calibration weights which have their own weight/mass printed on them, and which you can use to weigh and then tweak the scale’s readout so that it reads what it should for a particular calibration weight. If a scale doesn’t have a tweaking option, keep a reference card handy so you know that the scale is X grams or X ounces off when it measures the amount your calibration weight says it should be. Also note that calibration errors can very well be non-linear, so if you have 5 calibration weights, use them all and see if you can get a bead on an average error or even, if geeky, plot a curve and extrapolate! If you want to cheap out and not purchase calibration weights, you can approximate by using water. 1 ml is 1 gram. 1 liter is 1 kilogram, roughly, at room temperature. But remember, if you go this way, your volume measurement must be as precise and accurate as possible.
Calibration is harder with thermometers and ovens. To calibrate a thermometer, first you need to make sure you’re at sea level and that you’re using distilled water. And that you’re using distilled water for the ice. But with these three things you can calibrate a thermometer to boiling and freezing. For boiling, boil the distilled water and make several measurements with the thermometer(s) you’re calibrating. The thermometer should read 100° C or 212° F. If not, mark it down. Some people actually use a permanent marker on the thermometer’s scale itself. Some folks write it down on a note stuck on the cabinet door. Now mix distilled water and distilled water ice cubes and measure that. The thermometer should read 0° C or 32° F. Again, note down any differences from expectations and/or make permanent marks on your thermometer so you can rely on a better measurement next time.
- There’s a science/mathematical procedure to carrying errors through scientific calculations so we know how correct our findings are, but not everyone uses or understands it well enough to use it and interpret it properly.
- Precision is the tightness of a cluster of measurements of the same thing around an average (which may not be anywhere near the true value, depending on accuracy).
- Accuracy is the closeness of the average of many measurements to objective truth.
- To fix accuracy issues, you can calibrate.
- To fix precision issues, you usually have to spend lots of money, or do a lot of calibration to test out the precision of different instruments.
- It can also help to develop and consistently use the best way of measuring (which usually increases precision).
- When doing measurements, truth is very subjective and contextual, depending a great deal on how well you manage observation and calculation errors.