A useful difference between clocks

A clock can be consistent without displaying the correct time. Imagine that a station clock gains exactly two minutes every day. Its readings are inaccurate, but its error follows a predictable pattern. A technician who knows that pattern can estimate the correct time from a reading, provided the clock was set correctly at a known starting point.

[A] Now consider a second clock whose readings sometimes run fast and sometimes slow by unpredictable amounts. [B] Even if its average error over a month is close to zero, an individual reading may be unreliable. [C] An average can therefore hide variation that matters to a person trying to catch a train. [D] For that person, the quality of each reading is more important than the balance of errors across many days.

The distinction is useful beyond timekeeping. When researchers evaluate a measuring instrument, they investigate both systematic error and variation between readings. Repeated measurements may reveal a stable bias that can be corrected, while irregular fluctuations may require better equipment or improved procedures. Reporting only an average leaves out information needed to judge whether a measurement is suitable for a particular decision.