A scale can be perfectly consistent and still wrong every single time
In everyday speech accuracy and precision mean the same thing, but scientists pull them apart. Accuracy is how near your readings land to the true value; precision is how tightly they cluster together. A flawed instrument can give beautifully consistent results that all miss the mark by the same amount.
Precision reflects random error, the scatter you would measure as a standard deviation, and it ties to how repeatable and reproducible results are. Accuracy, in its older sense, reflects systematic error, a steady bias that shifts the average away from reality. Because the two are independent, a data set can be accurate, precise, both or neither. Statisticians often prefer the plainer words bias and variability.
Averaging more readings sharpens precision but cannot fix a built-in bias. If an experiment has a systematic flaw, a bigger sample just yields a tidier cluster around the wrong answer; removing the flaw improves accuracy while leaving the scatter unchanged. When readings are averaged, the precision of the mean equals the process's standard deviation divided by the square root of the number of readings, and the central limit theorem makes averages look more like a bell curve than single measurements do.
Standards bodies shifted the vocabulary in 1994 with the ISO 5725 series, later echoed in the 2008 International Vocabulary of Metrology. There, closeness of the mean to the true value is called trueness, and accuracy becomes the umbrella term covering both trueness and precision, so high accuracy demands both. The ISO documents avoid the word bias because it carries different meanings in medicine and law. Instruments are checked by repeatedly measuring traceable reference standards defined in the International System of Units and kept by bodies such as the US National Institute of Standards and Technology.
Other fields bend the terms. In numerical computing, precision means how many digits a number is stored with. Resolution is the smallest change an instrument can register. Scientists also signal uncertainty through significant figures: writing 843 metres implies a margin of half a metre, while 843.0 metres implies five centimetres.
Source: Accuracy and precision
