How to build a science fair display board that actually works.

The Constraint

Two metres. Five seconds.

At a national fair a judge may have thirty projects to get through in an afternoon — about six minutes each, most of it spent talking rather than reading. So the board has to answer three questions before anyone speaks. What did you ask? What did you find? Should I care?

The test is easy and slightly brutal. Print the draft, tape it up, walk two metres back, count to five, look away and say what the project is. Better: get someone who knows nothing about it — a sibling, a neighbour — to do it. Whatever they say back is what your board actually communicates.

Size & Rules

How big it is allowed to be.

The standard tri-fold sold in most stationers is about 91 cm tall and 122 cm wide open: a 61 cm centre panel with two 30 cm wings. That shape exists because it stands up on a table by itself, which matters more than you think in a hall with no wall space.

Regeneron ISEF limits the whole display — board, table items, everything — to roughly 122 cm wide, 76 cm deep front-to-back and 274 cm tall measured from the floor. In practice a standard tri-fold on a standard table is comfortably legal. Do not take those numbers from me as gospel: they get revised, regional fairs add their own restrictions, and the only version that counts is the current official rules at societyforscience.org.

Two things catch people out. Depth: a tall apparatus leaning forward, or a second table pushed to the front, breaks the 76 cm limit fast. Height: it is measured from the floor, not the tabletop, so 274 cm minus a 76 cm table leaves under two metres of usable space.

If your fair allows it, a single flat A0 poster (84 x 119 cm) often looks more professional — but it needs a wall or an easel, so check what the venue provides.

The Layout Map

Where everything actually goes.

People do not read a tri-fold in a spiral or a Z. They hit the title, drop to whatever the biggest image is, then read the left wing top-to-bottom, then the right wing. Design for that order instead of fighting it.

Notice what is missing. No abstract, no literature review, no raw data tables, no acknowledgements eating a quarter of a wing, no border of clip-art atoms. Leave roughly a third of the board as empty background, keep a consistent 3 to 4 cm gutter between blocks, and align their tops to an invisible grid. That much space feels wrong the first time and is the single change that most improves how a board reads.

Typography

Type that survives distance.

A rule of thumb borrowed from signage works well here: allow about one centimetre of capital-letter height for every metre you want text read from at a glance. Body text is different, because people step in to read it — but there is still a floor below which they will not bother.

Left-align every paragraph. Centred body text is the most common mistake on science fair boards, and it is a reading problem rather than a taste one: when every line starts at a different position, the eye has to hunt for the start of the next one instead of snapping back to a fixed edge. Centre titles and single lines only. Do not justify either — it opens rivers of white space at large sizes.

Two typefaces, maximum — one for headings, one for body, with a heavy weight for the title. Avoid anything condensed, script, or textured: at distance, letters need open shapes and clear spacing to stay distinguishable.

Contrast beats colour. Dark charcoal on white or very pale card is safest in a hall, where lighting is unpredictable and often yellow; white on dark looks striking on your laptop and frequently prints muddy. Use one accent colour, hold line length under about 60 characters, and set line spacing near 1.35. If a paragraph runs past four lines at 24 pt it belongs in the report — running the text through Hemingway Editor first will show you which sentences are doing too much.

Charts

Three to five charts, one hero.

The commonest charting failure is not an ugly chart, it is six equally sized ones. Equal size tells the reader everything mattered equally, which is never true. Decide which figure answers your question, make it dramatically bigger, and let the rest support it.

A worked example. You tested three filter materials against a control, five replicates each, measuring removal percentage. The hero chart is four bars with standard-deviation error bars, a y-axis starting at zero, axis labels at 18 pt or larger, and a headline above it reading “Material B removed 62% ± 4% — nearly three times the control” rather than “Figure 3: Results”. A judge who reads only that sentence and that chart has understood your project. Your supporting charts are then the same data across trials, showing the effect was stable, and removal against particle size, showing where the mechanism breaks down.

Export at 300 dpi or as vector PDF. A chart screenshotted from a spreadsheet looks soft at 30 cm wide and is the most visible sign of a rushed board. Never use a 3D bar chart — the perspective distorts the exact comparison you are trying to show.

Do not rely on colour alone to separate series: roughly one in twelve men has some form of colour vision deficiency, so there is a fair chance a judge does. Add patterns, direct labels or different marker shapes. ColorBrewer has colourblind-safe palettes built for this, and Datawrapper makes clean print-ready charts free in a browser.

One thing judges notice instantly: error bars. A bar chart with no indication of spread says you either ran one trial or did not think about variation. The statistics guide covers what to calculate and what those numbers actually mean.

Print & Mount

What to print, mount and laminate.

Print in panels, not one enormous sheet. A single A0 looks lovely until you spot a typo the night before. Build the board as A4 and A3 blocks, one per section, laid out at the right physical size — then a mistake costs one panel and forty cents.

Ask for matte paper around 160 to 200 gsm. Standard 80 gsm office paper ripples the moment adhesive touches it and shows every lump underneath. Matte also kills glare: under hall spotlights a glossy A3 panel becomes a mirror.

Laminate only the small things people handle — a setup photo, a data card, a QR card linking to your dataset. Those get touched by dozens of people and are otherwise tattered by lunchtime.

Getting There

The journey ruins more boards than the design does.

Make a sleeve. Two sheets of flattened cardboard box, a few centimetres larger than the folded board, taped along three edges. Slide the board in, tape the fourth, carry it upright with the fold at the bottom. Fifteen minutes of work, and the difference between arriving with a board and arriving with crushed corners. If you are flying, print the panels at home, carry them rolled in a poster tube and buy the tri-fold at the destination — unmounted panels travel far better than a mounted board.

The notebook matters more than students expect. An interested judge will ask to see it, and dated entries with failed runs and messy corrections are more convincing than any amount of polish on the board. What you may physically display is set by your fair’s safety rules — the safety and ethics guide covers the forms and approvals you need before you turn up with anything.

When It Goes Wrong

The failures you will recognise.

I have watched every one of these happen and caused a few myself. None are about intelligence. They all come from forgetting that the board is read at distance, in a hurry, by someone who did not do the work.

The worst I ever saw was a genuinely excellent project — careful controls, a clever measurement, months of work — on a board whose whole centre panel was a photograph of the student holding the apparatus, with the results in a 14 pt table in the corner of a wing. Judges stopped, looked, could not find the finding, and moved on. The science was the best in the room and the board never let anyone know. That is the failure worth fearing, and it is preventable in an afternoon.

Common Questions

What students ask about boards.