Zero, Tare and the Blank: Every Instrument Reads Something When It Should Read Nothing

Zero it closed, level, at working temperature, in the position it will be read in. An instrument zeroed cold and used warm is not zeroed, and one zeroed flat and used upright often is not either. Then check the zero AGAIN at the end. If it has moved, you know the readings drifted and roughly by how much; if it has not, you have positive evidence that they did not. That second check costs five seconds and is the difference between hoping and knowing. Some things cannot be re-zeroed and must be recorded instead. A steel rule with a worn end does not have a zero — measure from the 10 mm mark and subtract 10, which is the same trick as measuring everything from one origin.

Taring puts the container on the balance and sets that reading as the new zero, so the display shows only what you add. Weighing container-plus-contents and subtracting a remembered container weight gives the same arithmetic and a much worse result. Two reasons. The container's weight CHANGES — it picks up moisture, it keeps a film of the last thing in it, it is not the one you weighed last week. And subtracting two large numbers to get a small one throws away precision: 250.14 minus 248.09 has far fewer meaningful digits than either number looks like it has. So tare with the ACTUAL container, immediately before use, every time. If the balance has no tare, weigh the empty container in the same session rather than trusting a figure written on its side.

Take your procedure exactly as written, apply it to a sample you know contains none of what you are measuring, and see what number comes out. That number is your BLANK, and it belongs to the method rather than to the instrument. It catches what zeroing cannot: contamination in the reagents, colour in the glassware, absorbency in the filter paper, a balance pan that is not clean. A blank of zero is a strong statement that none of those is happening; a large blank tells you to fix the method before taking a single real reading. Run the blank in the SAME session as the samples, not once a month. Reagents go off, glassware gets used for something else, and a blank from last week is evidence about last week.

Zeroing fixes the bottom of the scale. It says nothing about whether 100 g reads as 100 g, and a balance can be perfectly zeroed and still 2% out at the top. So check it against a known value: a calibration weight, a gauge block, ice water and boiling water for a thermometer. Two points is the minimum — one near the bottom of the range you use and one near the top — because a single point cannot tell a shifted scale from a stretched one. Write the check down with its date. A calibration is a measurement like any other, it has its own uncertainty, and it stops being evidence after a while. The Celsius rung is this same idea done properly with two fixed points that nature supplies for free.