What does "artefact" mean in the context of science?

In science, an artefact usually means something that shows up in your observations or data because of the way you studied the subject, rather than because it’s really part of the subject. An air bubble that looks like a cell on a microscope slide is an artefact. So is a streak on a CT scan caused by a metal filling, or a “trend” in a dataset that comes only from the way the sample was chosen.

The word has a second, older meaning that you’ll meet in archaeology and anthropology: an object made or shaped by people, such as a stone tool or a pot.

Both meanings come from the same idea, something made by human skill. In the experimental sense, the “something made” is a feature that your own methods have created by accident.

Artefact or artifact?

They’re the same word with two spellings:

Artefact is the usual spelling in British English and in many Commonwealth countries.
Artifact is the usual spelling in American English, and it’s common in international scientific journals.

Both come from the Latin arte, “by skill”, and factum, “made”. Use whichever matches the style you’re writing in, and keep it consistent.

Artefacts in archaeology and anthropology

In archaeology, an artefact is a portable object made, modified or used by humans: tools, ornaments, pottery, coins. Archaeologists often contrast artefacts with:

Ecofacts: natural remains that tell us about past human life, such as animal bones, seeds, pollen and charcoal.
Features: things that can’t be moved without destroying them, such as hearths, post holes and walls.

Philosophers also discuss artefacts in this sense, asking what makes something an artefact rather than a natural object, and how we define an artefact’s function. That’s a rich field, but it’s not usually what a scientist means when they warn you about “an artefact in the data”.

Artefacts in experimental science

In laboratory and field science, an artefact is a false or distorted signal introduced by the process of observing, preparing, measuring or analysing. It looks like information about the subject, but it’s really information about your method.

Artefacts matter because they can be mistaken for real discoveries. A good scientist’s first question about a surprising result is often: “Could this be an artefact?”

Artefacts can come from almost any step:

Collecting the sample: crushing, contamination, delay before preserving.
Preparing it: chemical fixation, drying, freezing, cutting, staining.
The instrument: optical flaws, electrical noise, calibration errors.
The observer: expectation and suggestion, or seeing patterns in noise.
Processing and analysis: image compression, filtering, statistical choices.

Examples across fields

Microscopy

Microscopy is full of artefacts, which is why this site has a full guide to microscopy artifacts and how to avoid them. Common ones include:

Air bubbles in wet mounts, which look like round cells with thick dark outlines. Our guide to making a wet mount slide without air bubbles shows how to avoid them.
Shrinkage and cracks in tissue caused by fixatives and dehydration, which can create gaps that weren’t there in life.
Folds, tears and knife marks in thin sections.
Stain precipitate: crystals or granules of dye that look like bacteria or cell granules.
Debris and contaminants such as fibres, starch grains from gloves and dust on the coverslip.
Uneven smears: in a poorly made blood film, red cells can look misshapen simply because the smear was too thick or dried too slowly. Our guide to making a blood smear slide covers this.

Medical imaging

Motion artefacts: blurring or “ghost” copies in MRI and CT when the patient moves or breathes.
Metal artefacts: bright and dark streaks radiating from dental fillings, implants or surgical clips on CT.
Ultrasound artefacts: shadows, reverberations and mirror images that don’t correspond to real structures. Sonographers learn to recognise them, and some are even useful clues.

Signals and recordings

ECG and EEG artefacts: muscle twitching, electrode movement and electrical interference from mains power can produce waves that mimic a heart or brain abnormality.
Audio artefacts: clicks, hiss, distortion and compression noise in recordings.

Statistics and data

Statistical artefacts: patterns created by how data were collected or analysed. For example, a correlation may appear only because of who was included in a study (selection bias). An apparent improvement may be regression to the mean. Or a trend may appear simply because the way something was measured changed over time.

Digital images

Compression artefacts: blocky squares and halos around edges in heavily compressed JPEG images.
Aliasing: moiré patterns and jagged edges when fine detail is sampled too coarsely, for example when photographing a screen or fine fabric.
Processing artefacts: halos from over-sharpening, and false detail from aggressive noise reduction or deconvolution.

What is a visual artefact?

A visual artefact is any artefact you can see in an image: a feature of the picture that isn’t a feature of the thing pictured. Lens flare, chromatic fringes, dust spots on a camera sensor, compression blocks and screen tearing in video games are all visual artefacts. In microscopy, the halo around objects in phase contrast is a visual artefact of the technique. Experienced users learn to look past it.

How scientists tell an artefact from a real finding

Researchers use several standard checks:

Controls. Run the same procedure on a sample known not to contain the thing you’re looking for. If the feature still appears, your method is probably creating it. In staining, for example, labs often run a slide with a key reagent left out.
Change the method. Look at the same subject a different way: a different fixative, stain, instrument or analysis. Real structures tend to show up whichever reasonable method you use. Artefacts often change or disappear.
Repeat it. Ideally other people, in other labs, with other equipment. An artefact tied to one setup usually fails to replicate.
Look at living or unprocessed material where possible. If a structure only appears after harsh preparation, be suspicious.
Know the usual suspects. Every field keeps reference guides to common artefacts, so newcomers learn to recognise bubbles, folds, precipitates and electrical noise.
Blind the observer. People who don’t know which sample is which are less likely to “see” what they expect.

Artefacts that fooled scientists

A few well-known cases show why the word carries such weight.

Mesosomes. From the 1950s, electron micrographs of bacteria showed folded membrane structures called mesosomes. For years they were described as real cell organelles. Later work with gentler preparation methods, such as rapid freezing, largely failed to find them. Most microbiologists now regard mesosomes as an artefact of the chemical fixation used to prepare samples.
“Faster-than-light” neutrinos. In 2011 the OPERA experiment reported neutrinos apparently travelling slightly faster than light. The team traced the result to equipment problems, including a badly connected fibre-optic cable affecting the timing. The result was withdrawn.
Polywater. In the 1960s and early 1970s, some researchers reported a new, viscous form of water. It turned out to be ordinary water contaminated by impurities from the glassware and other sources.

In each case the scientific process worked eventually. Other researchers checked the results with different methods and found the artefact. That’s why the first habit to develop at the microscope is the same one professionals rely on: before you believe something new, ask whether your own preparation could have put it there.