AbilityBenchFree, 4 sweeps at 2.5° per second
Blind spot test that reports where your optic disc actually sits
Stretch a card outline to the width of a real bank card, type how far your eye is from the screen, and the page can convert pixels into degrees of visual angle — which is the whole difference between a demonstration and a measurement. A dot then crosses the screen at 2.5° per second while you hold fixation with one eye covered; you mark where it vanishes and where it returns, twice outward and twice inward, and the page reports the middle of your optic disc, its width and its height. It is free, it asks for no account, and both numbers you type stay in the tab.
- 100% free
- No signup
- 4 sweeps per eye
- 0.7° dot
- Needs a bank card
Every eye has a patch of retina with no photoreceptors on it, where the optic nerve leaves for the brain. This page finds where yours is in degrees of visual angle instead of just telling you it exists: a dot crosses the screen at 2.5° per second, and you mark the place where it goes and the place where it comes back, twice in each direction.
The track above reaches 0.0° sideways and 0.0° up or down at this calibration. A disc can sit further out than that, so if the dot never disappears, move the screen closer: at 25 cm this same box would cover 19°, because the pixels stay where they are while each degree of your field takes fewer of them. Set the card and the distance again afterwards.
How to measure your own blind spot
Calibrate the screen, cover one eye, and mark the two edges twice in each direction.
Set the card outline and measure your distance
Hold any bank card, library card or driving license against the outline on screen and drag the slider until the two are exactly the same width — every card in the world is 85.6 mm across, so that one adjustment tells the page how big a pixel is on your particular panel. Then measure from your eye to the glass with a tape or a piece of string and type the centimeters in. Guessing this is the single largest error you can introduce: at 55 cm, being 10 cm out moves every degree below by about a fifth.
Cover one eye and put your gaze on the cross
Use a palm rather than a squint, because a half-closed eye still contributes. Look at the fixation cross and keep looking at it — the dot is not what you are watching, it is what you are noticing in the corner. If your eyes drift toward the dot the edges land wherever your gaze happened to be, and that error has no sign, so it cannot be averaged out the way the press lag can.
Mark the vanish and the return, both ways
Press the space bar the instant the dot disappears and again the instant it reappears. Two sweeps run outward from 4° and two run back inward from beyond your outer edge, because a late press pushes an outward edge too far out and an inward edge too far in by exactly the same amount. The page averages the pair to remove that lag, and prints half their disagreement as the lag itself.
Technical specifications
| What you have to supply | Two things a browser cannot see: the card width in pixels, from an ISO/IEC 7810 ID-1 card at 85.6 mm, and the eye-to-panel distance in centimeters. Their product is pixels per degree, and every angle on the page is drawn from it |
|---|---|
| Stimulus | A 0.7° black dot on a white field. Well under the disc's own width on purpose — a target wider than about 5° can never vanish whole, which is why other pages use a dot rather than a shape |
| Sweep speed and track | 2.5° per second, along a track set 1.5° below the fixation cross rather than level with it, because the disc projects slightly under the horizontal meridian |
| Sweeps per run | 4 across: two outward starting at 4° of eccentricity, then two inward starting 3° beyond the outer edge the first pair found. An inward sweep has to begin outside the disc or the dot is already missing when it sets off |
| Optional vertical run | 2 more sweeps, down and up through ±6° at the eccentricity the sideways run established, which is what turns two edges into an oval |
| Marks the page throws away | Any mark inside the first 0.6° of travel, because a dot that has barely moved has crossed nothing; and every mark in a sweep during which the tab lost focus, which is discarded and re-run rather than kept |
| Angular resolution | One frame of a 2.5°/s sweep moves the dot 0.042° at 60 Hz, so display timing is worth about a fortieth of a degree here. The 0.7° dot is the real floor: an edge is where the middle of the dot was, and no dot resolves a boundary finer than itself |
| The reference beside your figures | 12-18 degrees temporal to fixation, ≈ 5 degrees across and ≈ 7 degrees tall, sitting about ≈ 1.5 degrees below the horizontal meridian — Position of the optic disc, the one patch of retina with no photoreceptors, as mapped in routine perimetry. It is a span, not a distribution, so no percentile is computed from it |
Frequently asked questions
Why does this need a bank card and a tape measure?
Because a degree of visual angle is not a number of pixels until somebody says how big a pixel is and how far away you are sitting. A browser can read your window in CSS pixels and nothing else: it does not know your panel's physical size, its scaling factor, or whether your face is 40 cm or 80 cm from it. Every page that shows you a dot and announces your blind spot is 15° across has skipped both unknowns and is reporting the number it was told to report. The card fixes the first — 85.6 mm is guaranteed by ISO/IEC 7810 for every card in your wallet — and the tape fixes the second.
The dot never disappeared. What went wrong?
Almost always one of four things, in this order. Your gaze followed the dot, which keeps it on working retina the whole way across; the covered eye was the one being tested, so the open eye simply saw it; you are close enough to the screen that the disc lies beyond the edge of the window, which is why the page offers full screen and reports the eccentricity it can reach; or you swept along the horizontal meridian rather than the track set 1.5° below it, which is the reason the track is dropped in the first place. Somebody with a very narrow window and a short viewing distance may genuinely be unable to reach 15° on screen, and the page says so instead of pretending.
Why does the dot move by itself instead of letting me drag it?
Because a dot you control returns your own prediction rather than a measurement. Nudging a marker with an arrow key means choosing when to stop, and people stop where they expected the edge to be. A constant sweep takes that away and substitutes a different error — your press lag — which has the advantage of being signed: it pushes an outward edge outward and an inward edge inward. Two sweeps each way therefore cancel it in the mean, and half the disagreement between the directions is a measurement of the lag itself, which the page prints in milliseconds.
Is my blind spot the right size?
There is no right size, and that is why the page does not grade you. The optic disc is an anatomical structure whose projection varies across the published span between healthy eyes, and the figure you get also carries your card setting, your distance, your fixation stability and the width of the dot. Read your two eyes against each other instead: run the left and then the right, and the interesting comparison is whether they come out roughly symmetrical, not whether either one landed on a number from a textbook.
Why is my vertical measurement bigger than my horizontal one?
Because the optic disc is taller than it is wide, so the oval it projects into your visual field is too. The page measures the two axes in separate runs — sideways first, because the vertical sweep has to be run at an eccentricity the sideways run has already found — and the difference between them is the expected result rather than an inconsistency. Expect the vertical extent to come out the larger of the two by a couple of degrees.
Why is there no black hole in my vision if the gap is that big?
Because the gap is not dark, it is absent, and the visual system fills it with whatever the surrounding retina reports. Darkness would itself be information; a region with no photoreceptors sends nothing at all, and what arrives in your experience is an interpolation. The page demonstrates this deliberately after the measurement, by drawing a line with a break in it exactly as wide as your own disc and exactly where your own sweeps put it — cover the other eye and the line reads as continuous, which is the completion doing its work at a size you established yourself.
Can this find something wrong with my visual field?
No, and it is not close enough to be worth using that way. This page probes one small region that everybody has, along one track, with your own hand as the recorder and no way to know where your eye is pointing. A visual field examination puts you at a bowl of controlled luminance, occludes one eye properly, monitors fixation, and probes dozens of positions out to the edge of the field precisely because the losses that matter are the ones you cannot notice yourself. If you have a reason to be worried about a gap in your vision, that examination is the thing to book, and nothing here substitutes for it.
Mariotte's hole, and the two numbers a browser has to be told
The blind spot has been a known object since Édme Mariotte published it in 1668, and its cause is plain anatomy: the axons of the retinal ganglion cells have to leave the eye somewhere, and where they do there is no room for photoreceptors. That patch of nothing projects into the visual field on the temporal side, because the disc sits on the nasal side of the retina and the eye inverts. Everything the page reports is arithmetic on top of that: the inner edge where a moving dot vanishes, the outer edge where it comes back, their midpoint as the position of the disc and their difference as its extent. There is no threshold being estimated. The dot is far above contrast threshold the whole way across, and the only reason it goes is that there is nothing underneath it.
What separates a measurement from a demonstration here is the conversion from screen to angle, and it is the step almost every online version skips. A page that draws a dot and announces an eccentricity has quietly assumed a pixel size and a viewing distance, so its number is a property of the assumption. This one asks for both and shows the multiplication: a card outline stretched to 85.6 mm gives pixels per millimeter, the distance gives millimeters per degree, and the product is what the sweeps are drawn in. That same pair of inputs is what the peripheral vision test needs, so the two pages share one calibration panel rather than asking for a card in two different ways. Its sibling in this group needs neither: the dominant eye test happens in the room rather than on the panel, because a sighting test needs a near object and a far one and a flat screen only has one distance.
The other thing worth knowing before you read your own figure is that the page is measuring your finger as well as your retina, and it does something about it. A press arrives after the event that caused it, so an outward sweep records every edge too far out and an inward sweep records it too far in — the same lag, opposite sign. Averaging the two directions removes it and halving their difference recovers it, which is the one internal consistency check this method offers and the reason the run is four sweeps rather than one. Geometry is what a browser can pin down honestly; where it cannot, the pages on this site say so instead of inventing a figure — the contrast sensitivity test cannot know your panel’s luminance and the color perception test cannot know its white point, while a length judgment in the optical illusion test survives both unknowns. The measurements that need no display calibration at all are the ones made of items rather than geometry, which is the family the Mensa-style IQ test and the Wonderlic practice test belong to — they have the opposite problem, where the difficulty is the content rather than the panel.
This is a measurement exercise, not a clinical assessment. It reports what you did on this page against a stated reference and nothing more — it cannot establish a scotoma, glaucoma or any other gap in your visual field. Only a qualified professional, working with more than a browser, can make that judgment.
Every eye has this gap, in the same place, for the same structural reason, and the variation between healthy people covers the whole published span. A run that lands anywhere inside it is not evidence of anything, and a run that lands outside it is far more likely to be a mistyped viewing distance than a finding.
A reaction time here is the interval between the frame that painted the stimulus and the timestamp the browser attached to your key, both read from the same monotonic clock. What neither can see is the display pipeline behind it, so on a 60 Hz screen roughly 16 ms of every figure below is the machine rather than you. That is the timing floor: two numbers closer together than that are the same number, and this page reports no precision it cannot support.
Only one figure on this page is in milliseconds, and it is not a reaction time: it is the press lag recovered from the disagreement between the outward and inward sweeps. Here the display floor arrives as a distance rather than a duration — a 60 Hz frame moves the dot 0.042° — so it is swamped by the 0.7° width of the dot itself, and the degrees are quoted to a tenth for that reason rather than out of caution.
Where your card width and your viewing distance go
Every number on this page is worked out by JavaScript running in the tab you are reading it in. Your answers, your reaction times and your score are never uploaded, logged or kept — which is also why the test carries on working after you disconnect from the network, and why nothing here can be held back behind an email address.
The two calibration numbers are held in this tab as a card width in pixels and a distance in centimeters, used to draw the sweep, and dropped when you close the page — they are not written to local storage, so a reload asks again. Nothing recovers a finished run either, which is what the copy button is for: it puts the edges, the center, the extent, the press lag and the calibration you used on your own clipboard as plain text.