The Camera Obscura: An Image Without Chemistry

Black out a room, cut a clean hole in the blackout, and the wall behind you fills with the street outside: in colour, moving, and upside down. No lens, no glass, no chemistry, no electricity. The picture is there before anything records it, and by the end of this page you will have built two instruments that show it, predicted the size of the image before measuring it, and found out for yourself what you give up every time you make the hole bigger.

Purpose

To separate the two halves of photography by doing the first half on its own. Forming an image is optics; keeping it is chemistry. Everything in the rest of Part I is about the second half failing, and none of it makes sense until you have stood in front of the first half working.

Concretely, you build two cameras obscura at different scales, test them for light-tightness, and make three measurements: the height of the projected image at two screen distances against the height similar triangles predict, and the brightness and detail given by three interchangeable apertures.

The box you build here is not a demonstration you throw away. It becomes the body of the pinhole camera in Part VI, which adds a light-tight back, a shutter and a film or paper holder. That constrains this build in two ways, and both are called out where they arise: the box must be rigid enough to survive a year on a shelf, and the aperture must be a small replaceable plate rather than a hole punched in the box itself.

Learning objectives

By the end of this page you should be able to:

• state what the rectilinear propagation of light is, and explain the inverted, laterally reversed pinhole image as a consequence of it rather than as a curiosity;
• predict the height of a projected image from the object’s height and the two distances, using similar triangles, and say by how much your measurement missed;
• explain, in terms of the ratio of hole diameter to screen distance, why enlarging the hole brightens and blurs the image by related amounts, and why a room camera and a box camera can be the same camera at different scales;
• say why there is a best hole size without yet being able to calculate it;
• test an enclosure for light leaks in a way that does not depend on how good you think your eyes are;
• describe how the camera obscura reached the eighteenth-century artist’s box, and distinguish what the historical record shows from what later writers inferred from it.

Prerequisites

None, beyond a table to work on. Nothing here is chemical, electrical or optical in a way that assumes prior study; if you can measure a distance and cut a straight line you have what you need. Read the part overview first if you have not.

It helps, but is not required, to have read Silver salts and light, because the last section here hands over to it: the image you are about to make is the one that eighteenth- and nineteenth-century experimenters could see perfectly well and could not hold on to.

Safety classification

Level A, on the criteria of the classification rubric:

• Substances and quantities. Nothing is weighed, dissolved, diluted, poured or mixed. The materials are card, tape, foil, fabric and, if you choose, a magnifying glass.
• Energy. Nothing is heated, no flame is used and no ultraviolet lamp is switched on. The only energy source is daylight, and the one genuinely dangerous thing you can do with daylight and an optical instrument is dealt with in Hazards below.
• Electrical. Nothing is wired or plugged in. A torch is the only powered item, and a small battery LED torch is what the build asks for.
• Waste. Card offcuts, tape and a scrap of foil.

What is not a hazard here, and why. There is no chemical hazard on this page at all, because there is no chemical on it: nothing is opened, decanted or brought into contact with skin. That is a statement about this specific build and not about photography, and it will not survive Part II, where the first silver nitrate solution arrives and brings nitrile gloves and splash goggles with it. Nor is there an electrical hazard, because there is no mains circuit; the enlarger, the ultraviolet unit and the timer in later parts each carry their own assessment. Ventilation is not among the controls here either, for the reason given below. What is left is a knife, a window and the sun, and those three get the whole of the next section.

Hazards

Never look at the sun. Not with your eye, not through the pinhole, not through the lens, not through a phone camera, and not through any filter improvised for the purpose. NASA’s guidance is unambiguous on both counts: it is not safe to look directly at the sun without specialised solar eye protection, and viewing the bright sun through a camera lens, binoculars or a telescope without a special-purpose solar filter secured over the front of the optics will instantly cause severe eye injury. The camera obscura is the answer to that problem rather than an instance of it, and the distinction is the whole point of the instrument:

⚠Projection, not viewing

A pinhole projector works with the sun behind you. Light passes through the hole and falls on a screen, and you look at the screen. NASA describes exactly this arrangement — a small hole, an image of the sun thrown onto a nearby surface, the observer with their back to the sun — as an indirect method that does not involve looking at the sun, and follows it immediately with the one prohibition that makes it work: do not look at the sun through the pinhole. Build the instrument so that the only way to see the image is to look at the screen, and this hazard is designed out rather than managed.

A lens concentrates sunlight, and that is a burn and a fire. The optional lens upgrade collects the light falling on a disc tens of millimetres across and brings it to a spot about a millimetre wide. Never point the lens camera at the sun, never leave it sitting on a windowsill pointing anywhere near it, and cap or cover the lens whenever the camera is not in use. The same physics that makes the image brighter makes the spot hotter.

Solar ultraviolet on skin, over a long session. The measurements ask you to stand outside with a box on a wall for as long as it takes, and the failure mode is not a dramatic one: it is having been out for ninety minutes without noticing. NASA states the control for anyone in direct sun for hours in one line — sunscreen, a hat and protective clothing — and ICNIRP supplies the field test that needs no equipment: when your shadow is shorter than you are, the ultraviolet is strong. The fuller treatment of solar ultraviolet, including what happens under broken cloud and how much of it comes through window glass, belongs to the unfixed photogram, which is the page in this part whose whole exposure happens outdoors.

Cuts. Card, foil and blackout material all need cutting, and a fresh craft-knife blade parts skin more easily than it parts corrugated card. HSE’s guidance on knives is written for kitchens, so take from it the parts that transfer: keep the blade sharp, because a blunt one needs force and force is what slips; cut on a stable surface; store the knife securely the moment you put it down, rather than leaving it loose on the bench where it can be knocked off; and never try to catch a falling knife. To that add the two rules that belong to craft cutting specifically — cut on a mat, with the stroke moving away from the hand holding the work, and retract or sheathe the blade every single time you set the knife down, including the times you are about to pick it straight back up.

A blacked-out room you have to move around in. Clear the floor before you black out the window, not after. Know where the door is. Keep a small LED torch in your pocket, and use that rather than a candle or a filament lamp: blackout fabric, bin liners and card are all combustible, so the control here is elimination — nothing in this build needs a heat source, so no heat source goes near the blackout. If there is a radiator or a heater under the window you intend to use, turn it off and leave a gap between it and the material rather than draping over it.

Working at a window. Keep both feet on the floor. If the top of the window is out of reach, black out only the lower half, or choose a different window; a camera obscura works just as well through a small aperture low down. Standing on a chair to reach a window is not part of this build.

Required PPE

Two kinds, neither of them chemical.

Cut protection, at the bench. A self-healing cutting mat and a steel rule are protective equipment on this page rather than conveniences, and it is worth being clear why. A plastic or wooden rule lets a sharp blade ride up over its edge and into the hand holding it; a steel one stops the blade. The mat holds the work still and keeps the blade from skating on a hard surface. Buy both before you buy anything else on the parts list.

Sun protection, outdoors. A hat, covering clothing and sunscreen, per NASA’s guidance for people spending hours in direct sun, plus sunglasses that block ultraviolet. ICNIRP’s figures are worth knowing because they argue for the combination rather than for either item: sunglasses alone give an ocular exposure ratio of about 0.5, and sunglasses worn with a brimmed hat about 0.02, and ICNIRP also records that most summer clothing gives a protection factor above 10.

On gloves. There is no substance on this page to keep off your skin, and what the work actually needs is clean dry hands, so that tape holds and foil does not tear as you pierce it. Gloves and eye protection become standing equipment in Part II, where the first solution is poured.

Ventilation

There is no airborne contaminant to control on this page. No solvent is opened, nothing is heated, nothing is sanded and no paint need be a spray, so ventilation does not appear among the controls in the hazard list above. It appears here for a different reason: a small room sealed with blackout material warms up and goes stale surprisingly fast, and an uncomfortable observer stops observing. Open the door and the blackout between sessions, and never black out a room whose door will not open from the inside.

Parts

Two builds, and you need both. Most of this is already in the house.

Build 1 — the room camera obscura

Item	Quantity	Notes
Opaque blackout material	enough to cover one window with 100 mm spare all round	Blackout curtain lining, heavy black polythene, or several thicknesses of cardboard. It must pass no light at all: hold a candidate up to a bright window and look.
Gaffer tape or wide masking tape	1 roll	Gaffer holds on painted plaster; masking tape is kinder to the paint and needs more of it.
Stiff black card	1 sheet, about A4	The aperture plate for the room build.
Aluminium kitchen foil	a piece about 100 × 100 mm	Heavy-duty foil is easier to pierce cleanly than the thin sort.
Plain white card or a pale sheet	1, as large as you can manage	Optional. A white surface gives a brighter image than patterned wallpaper.

Build 2 — the box camera obscura

Item	Quantity	Notes
Rigid cardboard box	1, about 250–350 mm on its longest side	Double-walled if possible. It has to last until Part VI. A shoebox works; a stationery box works better.
Matt black paint or black paper	enough to line the inside	Interior reflections are the enemy of contrast. Black poster paint, black acrylic or black sugar paper all serve.
Tracing paper or greaseproof paper	1 sheet	The screen.
Aluminium foil	3 pieces, about 40 × 40 mm	One per aperture plate.
Stiff black card	1 sheet, about A5	Aperture plate frames and the light baffle.
Black gaffer tape	1 roll	Also the light seal around every joint.
A reading glass or magnifier	1, optional	For the lens upgrade. Anything from about +2 to +5 dioptres.

Tools

Craft knife and a self-healing cutting mat; a steel rule; scissors; a sewing needle and a fine pin; a 300 mm ruler; a tape measure that reaches at least 5 m; a soft pencil; a notebook; a small LED torch; and a phone, only as a camera for recording what you see.

A note on the needle. The three apertures are pierced, not drilled or punched. A sewing needle pressed slowly through foil laid on a scrap of card makes a rounder hole with cleaner edges than any hand punch, and a fine pin makes a smaller one. Turn the needle as you press, stop as soon as the point is through, then rub the back of the foil flat on the mat with your fingernail to remove the burr. A ragged hole is not a failed hole — it is a measurement waiting to be made about why sharpness dropped.

Estimated cost

Cost band £, and it can be nearly nothing. Blackout material and gaffer tape are the only likely purchases; a bin liner and masking tape substitute for both. The cutting mat and steel rule are the one place worth spending, because they are the safety equipment and they will be in use for the rest of the course. Numbers, dated, are in the laboratory planner rather than here.

Estimated consumables cost

The band above covers the whole build. What is consumed by it is smaller and almost all of it is household: the blackout, the tape and the card are used up, and the knife, the mat, the rule and the needles are not. Nothing here is photographic, which is why the planner has no dated figure for any of it.

Consumed	This session	Sourced price	Cost this session
Blackout material for one window	about 1.5 m²	None. card-and-paper-stock carries a cost band and no dated figure	—
Gaffer, masking and black photographic tape	about one roll of each across both builds	None. tape-and-adhesives carries a cost band and no dated figure	—
Rigid box, aluminium foil and stiff black card	one box, four foil squares, two card sheets	None. box-and-foil carries a cost band and no dated figure	—
Tracing or greaseproof paper	one sheet, the screen	None. card-and-paper-stock carries a cost band and no dated figure	—
Matt black paint or black paper	enough to line one box interior	None. matt-black-paint carries a cost band and no dated figure	—
Craft-knife blades	one or two; cardboard blunts them fast	None. A named price gap: a self-healing cutting mat, steel rule and craft knife	—

Not one row in that table has a sourced price, so this session has no consumables figure at all — only a list of what it consumes. A priced entry is a dated range to plan against, never a quotation.

Kept rather than consumed, and therefore not in the table: the cutting mat, the steel rule, the knife handle, the needles and pin, the rules, the tape measure and the torch. Part I calls the mat and the rule protective equipment rather than conveniences, so they belong to the band above and to Stage 0 of the planner, not to the cost of a session.

How a hole makes a picture

Light in a uniform medium travels in straight lines. That single fact — rectilinear propagation — is the whole mechanism, and every property of the image follows from it.

Consider one point on the scene outside: the top of a doorway across the road. It scatters light in every direction. Almost all of that light hits your blackout and stops. A narrow cone of it passes through the hole and lands on the screen as a small patch. Now take a second point, the bottom of the same doorway: it too sends a narrow cone through the hole, and that cone lands somewhere else. Because the rays travel in straight lines and all of them squeeze through one small opening, they have to cross there. Light arriving from above the hole continues downwards; light from below continues upwards; light from the left continues to the right.

One hole, straight rays, and an image that has to be upside down

Nothing in the drawing is optional. Straight rays plus one small opening force the crossing, and the crossing forces both the inversion and the proportion.

The map this produces is one-to-one: one point on the scene, one small patch on the screen, which is what makes the result a picture rather than a glow. And the inversion is worth pausing on, because “inverted and laterally reversed” sounds like two faults and is really one. A mirror reverses left and right and leaves up and down alone; a pinhole turns the picture through half a revolution, which reverses both at once. Stand the box on its side and the image stands on its side with it. You can test the whole claim in ten seconds with a sheet of paper carrying a large letter F, held up in front of the hole.

The size of the image, before you measure it

Two triangles share their apex at the hole. On the outside, the object of height h_o at distance u; on the inside, the image of height h_i at distance v. The rays are straight, so the triangles are similar, so the ratios of their sides are equal:

h_i / h_o = v / u

Image height from similar triangles

h_i is the height of the image on the screen, h_o the true height of the thing you are looking at, v the distance from the hole to the screen, and u the distance from the hole to the object. Any consistent unit works, because only the ratio matters.

🧮Worked example: predict, then measure

A doorway 2.0 m high stands 10 m from your window. Your box camera has its screen 0.30 m behind the aperture. Rearranging,

h_i = h_o × v / u = 2.0 m × 0.30 m / 10 m = 0.060 m, or 60 mm.

Two predictions fall straight out of that expression and both are tested under Testing and calibration. Double the screen distance and the image doubles in height. Walk the camera to half the distance from the doorway and the image doubles again. Measure both. If your measured height is consistently a few per cent short of prediction, ask what you actually measured u from: the formula counts distances from the hole, not from the front of the box or the middle of the room.

The same relation sets the field of view, which is the reason a deep box shows you less than a shallow one. A screen of width w at distance v subtends a total angle of 2 arctan(w / 2v). A 200 mm screen 300 mm behind the hole spans about 37°, which is roughly what photographic convention calls a normal angle of view. Push the screen to 600 mm and it spans about 19°, a short telephoto, at a quarter of the brightness. Nothing about a pinhole has a focal length in the sense a lens does; the screen distance sets the magnification and the angle, and you choose it when you cut the box.

=The prediction most people get wrong

There is no focus. Slide the screen anywhere you like and the image never comes into or out of focus; it only changes size and brightness. That is the sharpest single difference between a hole and a lens, and it is worth checking deliberately in the box camera before you fit the lens, because after you fit the lens it becomes obvious in the other direction: the lens has exactly one screen distance that works for a distant subject, and everywhere else the image is mush.

Brightness against sharpness

Now enlarge the hole, and follow both consequences at once.

Sharpness. A point in the scene no longer maps to a point on the screen. It maps to a small patch, because the hole itself has width. Take the object point at distance u and the screen at distance v: the cone through a hole of diameter d spreads to a patch of diameter d(u + v) / u on the screen. For anything much further away than the screen is deep — which is nearly everything you will photograph — that reduces to d. The geometric blur is about the diameter of the hole itself. A 1 mm hole cannot resolve detail finer than about 1 mm on the screen, whatever you do to the rest of the camera.

Brightness. Each patch on the screen is lit through the hole and only through the hole, so the light it receives goes up with the hole’s area, as d², and falls off with the square of the screen distance, as 1/v². Double the diameter and the image is four times brighter. Double the screen distance and it is a quarter as bright.

Put the two together and something useful drops out. Brightness depends on d²/v², so on the ratio d/v. Relative sharpness — blur measured against the size of the image rather than in millimetres — also depends on d/v, because the image scales with v while the blur does not. So d/v is the single number that describes a pinhole camera, and its reciprocal is the quantity you will meet as the f-number in Part VI, written N = v/d. A ratio of 1/500 is f/500.

🧮Why the room and the box are the same camera

The box: a 0.6 mm hole, screen 300 mm away. d/v = 1/500.

The room: a 6 mm hole, wall 3 m away. d/v = 6/3000 = 1/500.

Identical. Both are working at about f/500, both give the same image brightness, and both blur the image by the same fraction of its own width — which is why the room camera does not look crude next to the box even though its hole is ten times wider. Scale the camera and scale the hole with it, and nothing changes. This is the argument that tells you what aperture to pierce for a camera of any size, and it costs no more than the two paragraphs above.

So why not simply keep shrinking the hole, since sharpness improves as d falls? Because one thing in the argument refuses to scale. Light passing through a small circular aperture spreads by diffraction, and OpenStax’s treatment states the crucial part plainly: the effect is most noticeable when the aperture is small. Diffraction spreads the patch by an angle that grows as the hole shrinks, so past some point every millimetre you take off the hole makes the image blurrier rather than sharper.

Lord Rayleigh set both effects side by side in 1891 in a note on pinhole photography, taking from Petzval an estimate of the blur that is the sum of two terms — the geometric shadow of the hole, which shrinks with the hole, and the diffraction spread, which grows as the hole shrinks — and deriving from it the size that makes the total least. Rayleigh himself objected that simply adding the two extreme cases is not a rigorous treatment, and worked the diffraction problem out properly. Part VI does the arithmetic and computes the best aperture for the box you are about to build. Here, the qualitative fact is enough, and it is a fact you can demonstrate this afternoon with three pieces of foil: as the hole shrinks the image gets sharper and dimmer, until it starts getting softer and dimmer, and that turning point is real.

✦Tip

This is why the three apertures below are roughly 0.3, 0.6 and 1.2 mm rather than a single “correct” hole. They are a starting bracket spanning a factor of four in diameter and sixteen in brightness, chosen so that a box a few hundred millimetres deep should show a visible ranking. They are a prediction, not a measurement: the course has not published its own measured optimum for this build, and your ranking is the data.

Two thousand years of the dark room

The instrument is old, its history is genuinely murky at the early end, and the murk is instructive: it is the same kind of murk that surrounds Schulze’s date and the first photograph, and it comes from the same cause — later writers copying a summary rather than reading the source.

When	Who	What the sources this course read actually say
Late 4th to mid 3rd century BC	The Mohist Canons	Record inquiries in geometry, mechanics, optics and economics. The dating is the Stanford Encyclopedia’s; the optical passages themselves are not described there, and this course has not read them.
4th century BC	The Aristotelian corpus	Eder writes that mention of pictures formed through a small aperture is found in the words of Aristotle, without naming the work.
c. 1038	Ibn al-Haytham	In an essay Eder gives as “On the Form of the Eclipse”, describes the crescent sun cast through a narrow round hole onto a plane, and what happens as the hole is widened. Eder spells the name Ibn al Haitam.
d. 1344	Levi ben Gerson	Used the camera obscura for eclipses of the sun and moon, as his predecessor had.
c. 1500	Leonardo da Vinci	The first clear description: images entering a small hole into a darkened room and appearing on the opposite wall, upside down. Written in mirror script and not printed for centuries.
1540	Erasmus Reinhold and pupils, among them Gemma Frisius	Observed a solar eclipse with a pinhole camera.
1550	Girolamo Cardano	De subtilitate, p. 107: proposes putting a glass disc in the aperture.
1553 and 1588	Giovanni Battista della Porta	Magiae naturalis: the first widely read description; a concave mirror in the first edition, a convex lens added in the second.
1568	Daniele Barbaro	La pratica della perspettiva, p. 192: the spectacle lens of a long-sighted man, and a diaphragm to improve the definition.
1575	Francesco Maurolico	Photismi de lumine et umbra: explains why the sun’s image is round through a square hole.
c. 1620	Johannes Kepler	A revolving tent with a convex lens in a tube, in which he traced landscapes with a pen — reported in Sir Henry Wotton’s letter to Bacon.
1665 or 1685	Johann Zahn	Oculus artificialis teledioptricus: a portable box with lenses in a tube and a slanting mirror throwing the image upward, which is the reflex camera.
1671	Athanasius Kircher	Ars magna lucis et umbrae: a camera large enough for the artist to climb into through a hatch in the floor and draw.

Three things in that table deserve more than a row, and two of its rows deserve a warning label.

Ibn al-Haytham’s account is an experiment, not an observation. He does not merely note the crescent; he states the condition under which it appears. The image of the sun shows this peculiarity, he writes, only when the hole is very small; enlarge it and the picture changes, and the change increases with the width; make the aperture very wide and the crescent disappears altogether, the patch of light taking the shape of the hole — round if the hole is round, square if it is square. That is the whole of the aperture argument from the section above, written a thousand years ago, and it is followed by a discussion of what happens as you vary the distance between the aperture and the wall. It is also, note, an instrument for looking at the sun without looking at the sun.

Barbaro’s diaphragm is the first stop. He put a lens in the hole and then deliberately made the opening smaller to sharpen the result, which is the trade you have just derived, being made by hand in Venice. Eder points out that Barbaro’s account was published twenty years before Porta described the lens as though it were his own secret, and that a comparison of Porta’s two editions by Liesegang shows the lens clause was inserted into the older sentence — Porta’s famous “something I have kept silent about until now” originally announced the concave mirror, not the lens.

⁇What the table could not settle

Two publication dates. Zahn’s Oculus artificialis: Eder gives the first edition as Würzburg 1665, and reinforces it by dating it four years before a publication of Boyle’s from 1669, while Wall’s Dictionary of Photography gives Würzburg 1685. Porta’s Magiae naturalis: Eder gives the first edition as 1553 and supports it by adding that Porta wrote it in his fifteenth year, which fits a birth in 1538; 1558 is the date more often quoted elsewhere. The table shows both in each case, because the course has not examined either book.

Two attributions. The Mohist Canons are widely quoted as containing a description of the inverted pinhole image. The Stanford Encyclopedia’s account of the Canons dates them to between the late fourth and mid third century BC and confirms that they record inquiries in optics, but gives no detail of the optical passages, and this course has not read them in translation. So the row above says what the Canons are and when, and does not quote what they say about pinholes. Likewise, Eder attributes the earliest mention of an image formed through a small aperture to Aristotle without naming the work; the passage usually cited for it is in the Problemata, whose attribution to Aristotle is itself disputed, and the course could not settle that from the sources it read. The row therefore says “the Aristotelian corpus” rather than “Aristotle”.

⁇Did the painters use one?

That some artists used optical aids is not seriously disputed; Mike Ware, writing about the prehistory of photography, records the recent art-historical re-evaluation of these instruments and points to Martin Kemp’s The Science of Art (1990), Philip Steadman’s Vermeer’s Camera (2001), Doug Nickel’s essay on drawing machines (1989) and David Hockney’s Secret Knowledge (2001).

What is disputed is who, and how much. The strong version — the Hockney–Falco thesis, that optical projection underlies a broad swathe of Western painting from the fifteenth century, and Steadman’s reconstruction of a camera obscura in Vermeer’s own room — is argued from the paintings rather than from documents that say so. The course has not read those arguments or the published responses to them and takes no position on either. Two things follow for you. First, when you trace your own projection under Further experiments, you are testing a hypothesis about what a tracing does and does not capture, not re-enacting a settled fact. Second, “Vermeer used a camera obscura” is a different kind of sentence from “Barbaro described a diaphragm in 1568”: one is an inference from the appearance of the finished work, the other is a page in a book with a date on it.

And here is why a course about chemistry spends its first practical page on optics. Every person in that table could see the picture. Not one of them could keep it. When Thomas Wedgwood finally tried, around 1800, he found the camera image too faint to act on his silver nitrate paper at all — the failure that the permanence page is built around. The first person to hold a camera image was Niépce, and the Harry Ransom Center’s account of the surviving plate says how he did it: not with a better camera, but by leaving the plate in one for several days.

Build

Work through Build 1 first. It is cruder, it takes twenty minutes, and it will teach you more about light-tightness than an hour of careful boxwork.

Build 1: the room camera obscura

1. Choose the room. You want a window facing something bright and, ideally, sunlit: a street, a garden, a wall with movement in front of it. The room wants a plain pale wall opposite that window, two to four metres away. Small rooms and bathrooms are often the best candidates, because there is less window to black out.
2. Clear the floor of everything you could trip over, and check the door opens from the inside.
3. Black out the window completely. Tape the material to the frame, then walk the edges with more tape. Light leaks live at corners and along the top edge. It does not have to be tidy; it has to be dark.
4. Wait. See Testing and calibration below before you cut the hole: the light-tightness test is done on a blackout with no aperture in it.
5. Make the aperture plate. Cut a 60 mm square hole in the middle of your black card. Tape a piece of foil over it, flat and taut. Pierce the foil once, cleanly, with a needle, and enlarge the hole to roughly 5–8 mm by rotating a pencil point through it, keeping it round.
6. Cut a matching hole in the blackout — smaller than the card — and tape the aperture plate over it from the inside, so the plate can be swapped without redoing the blackout.
7. Let your eyes adapt for at least ten minutes, then look at the wall opposite.

🛠If you see nothing

Three things account for almost every failure, in order of likelihood. Your eyes have not adapted; give it the full ten minutes with your back to the aperture. The room is not dark enough, and the leaked light is washing the image out; go back to step 3. Or the scene outside is not much brighter than the room, which is a matter of waiting for the sun to come round. A white sheet or a large piece of white card held against the far wall lifts a marginal image into a clear one.

Build 2: the box camera obscura

This is the one you keep, so build it as though it has to survive a year in a cupboard, because it does.

The box camera in section, with the aperture plate removed

1. 1Aperture window, about 30 mm square — cut in the end panel; it never carries the hole itself
2. 2Card frame, taped around the window — holds the plate flat and lets it be swapped
3. 3Foil aperture plate, S, M or L — labelled on the outward face; keep all three
4. 4Light flap — card taped along one edge, covering the joint
5. 5Matt black lining, every internal surface — including the lid; this is what gives the image its contrast
6. 6Tracing-paper screen, taped taut from inside — viewed from outside, shaded by your hands or a cloth
7. 7Screen distance v, foil plane to paper plane — measure it and write it on the box; Part VI needs it

Drawn to show the relationships, not to scale. The only dimension that has to be right is v, because every prediction on this page is made from it.

1. Choose and prepare the box. Rigid, dry, with a lid or a face you can open and close. Note its internal depth — that is your screen distance v, and you will need the number.

2. Black the inside. Paint or line every internal surface, including the lid. Untreated cardboard is a good diffuse reflector, and every stray bounce lands on your screen as veiling grey. This single step does more for image contrast than anything else in the build.

3. Cut the screen aperture. In one end of the box, cut a rectangular window about 100 × 80 mm, leaving a margin all round. Tape tracing paper over it from the inside, taut and flat. This is the ground glass, and you view the image on it from outside, in the shade of your hands or a cloth hood.

4. Cut the aperture window. In the opposite end, cut a square hole about 30 × 30 mm, centred.

5. Make three aperture plates. Cut three 50 × 50 mm squares of black card. In the centre of each, cut a 15 × 15 mm window and tape foil across it. Then pierce each one:

   • Plate S: one careful press of a fine pin, withdrawn as soon as the point is through. Roughly 0.3 mm.
   • Plate M: a sewing needle, pressed until the shank is through. Roughly 0.6 mm.
   • Plate L: the needle pressed through and then rotated to open the hole up. Roughly 1.2 mm.

   Label them S, M and L on the card, on the side that faces out.

6. Measure the holes rather than trusting the description. Hold each plate against a bright window beside a steel rule and photograph it with your phone at the closest distance that still focuses, with the rule in the same plane; then measure the hole against the rule’s millimetre marks in the photograph. Write the three numbers in your notebook. They are the x-axis of every result on this page, and “roughly 0.6 mm” is a description, not a measurement.

7. Fit the plates so they swap. Tape a card frame around the aperture window so that a plate slides in and is held flat, and mask the joint with a flap of card taped along one edge. Do not tape a plate down permanently: Part VI needs this to be interchangeable.

8. Seal everything else. Tape every seam, every corner and the lid. Then test as below.

🛠The one detail Part VI depends on

Keep the aperture as a separate plate over a window, never as a hole in the box. Part VI derives an optimum diameter for a box of your exact depth and asks you to pierce a plate to match it; a hole punched straight through the box would leave you rebuilding the camera to test a number.

Optional: adding a lens

A reading glass turns the box into the eighteenth-century artist’s instrument, and it changes the character of the image completely: far brighter, sharp only at one screen distance, and sharp only over part of the field.

1. Find its focal length first. In a dim room, hold the lens up and project the image of a distant bright window onto a sheet of white card. Move the card until the image is sharpest, and measure lens-to-card. That distance is the focal length. Write it down.
2. The lens only works if the focal length is close to your box’s internal depth. If the focal length is much shorter, the image forms in mid-air inside the box and never reaches the screen; if much longer, it never forms at all. Match the lens to the box, or build a second, shallower box for the lens.
3. Mount it on a plate like the aperture plates, centred over the window, taped so that no light passes around its rim.
4. Fit a card stop — a disc with a smaller central hole — over the front of the lens, and note what happens to the sharpness at the edges of the field and to the brightness. That is Barbaro’s 1568 diaphragm, and you now have both his result and his reason for it.
5. Cap the lens whenever the camera is not in use, and re-read the second paragraph of Hazards before you take it outside.

Testing and calibration

Light-tightness, tested rather than assumed

Do this on the room build before you cut the aperture, and on the box build before you make any measurement.

Kodak’s darkroom-design sheet gives the test in one sentence: stay in the room for five minutes with all the lights turned off, and if you still cannot see a sheet of white paper held against a dark background, the room passes. The reason it works is that your eyes are doing the measuring, and by five minutes they are far more sensitive than they were at the start. Kodak’s safelight publication puts a number on the same effect from the other direction: it takes at least ten minutes for your eyes to become fully adjusted to the dark. So five minutes is the test; ten minutes is what you wait before judging the image itself.

Seal what you find, progressively. Tape closes small leaks; a second layer of material closes large ones. Then repeat the five minutes, because sealing one leak often reveals a fainter one behind it.

For the box, the equivalent test is to take it into the blacked-out room with a torch, put the torch inside with the lid shut and every aperture plate removed and covered, and look for the seams glowing from outside. Anywhere light gets out, light gets in.

Measurement 1: image height against screen distance

The point of this measurement is to catch yourself being wrong in a way you can quantify.

1. Choose a target outside with a height you can actually measure or reliably look up: a doorway, a window, a gate, a parked car’s wheelbase held vertically. Measure or pace the distance u from the aperture to it.
2. Set the box screen at its normal distance v, and measure v from the plane of the foil to the plane of the tracing paper.
3. Predict h_i before you look, using h_o × v / u. Write the prediction in the notebook.
4. Look, and measure the image on the tracing paper with a ruler held against it.
5. Make a second box, or fit a card screen at a different depth, and repeat at a screen distance as near to twice the first as you can manage. Predict, then measure.

	Trial 1	Trial 2
Object height ho (m)
Object distance u (m)
Screen distance v (mm)
Predicted image height (mm)
Measured image height (mm)
Difference (mm, and %)

Agreement inside about 10% is a good result with a tape measure and a paced distance. A consistent one-sided error usually means u was measured from the wrong place, or that the “2.0 m” doorway is 2.1 m.

Measurement 2: the three apertures

One variable changes: the diameter of the hole. The scene, the box, the screen distance and the light must not.

1. Set the camera up on something solid, pointed at a scene with fine detail in it — brickwork, a fence, foliage against sky, lettering on a sign.
2. Fit plate S. Let your eyes settle for a few seconds. Record: how bright the image is on a scale of 1 to 5, the finest detail you can make out and name, and the overall impression in a phrase.
3. Repeat with M, then L, then S again. That last repeat is not a mistake — it tells you how much your eye drifted during the sequence, and it is the closest thing to a control this measurement has.

Plate	Measured hole diameter (mm)	Brightness rank, 1 = dimmest	Finest detail resolved	Notes
S
M
L
S repeat

📓Lab notebook

The confound on this particular measurement is your own retina. Over the three or four minutes the sequence takes, your eyes go on dark-adapting, so a plate looked at last will tend to seem brighter than the identical plate looked at first, and the effect is easily as large as the difference you are trying to measure. Repeating plate S at the end is the whole defence: if the repeat ranks differently from the original, your drift is bigger than your signal and the run has to be done again in a shorter time or in a fixed order both ways. Note also what the table asks for. It asks which detail you can name, because “I can read the third line of the sign” is a statement another person can check, whereas “quite sharp” is not.

Recording the image

Photograph the tracing-paper screen with the phone braced on something solid, in a dark room, with the screen filling the frame. Use whatever long-exposure or night mode the phone offers, and — this matters more than the mode — use the same settings, the same distance and the same room light for all three apertures, and write down what they were. A set of three photographs taken with automatic exposure tells you nothing about relative brightness, because the phone has silently undone the very thing you were measuring.

Troubleshooting

What you see	Most likely cause	What to do
Nothing at all on the wall or screen	Eyes not dark-adapted	Ten minutes, back to the aperture, no phone screen. This is the commonest cause by a wide margin.
A dim grey wash with no picture in it	Light leaking around the blackout or through the box seams	Repeat the five-minute test. Seal, then repeat again; the second leak hides behind the first.
An image, but flat and milky	Interior reflections	Line or paint the inside of the box matt black, including the lid. In a room, the pale objects near the aperture are the culprits; move them.
Bright but nothing is sharp	The hole is too large, or is ragged	Fit plate S. If it is still soft, look at your foil under a bright light: a torn hole with a burr scatters. Pierce a fresh plate.
Sharp but too dim to see	The hole is too small for the light available	Fit plate L, or wait for better light. Do not enlarge plate S — you need it as a data point.
The image is sharp in the middle and smeared at the edges	Expected with a simple lens, not with a hole	If the lens is fitted, this is field curvature and aberration, and stopping down with a card disc will reduce it. If no lens is fitted, the screen is not perpendicular to the axis.
A bright spot or streak across the image	A specular reflection off a shiny interior surface or the foil’s own edge	Black the offending surface. Check the inside face of the aperture plate is the black one.
The image moves when you touch the box	Nothing is wrong with the optics	Brace the camera. It matters far more once Part VI puts a several-second exposure behind that aperture.
The tracing paper has gone limp and wrinkled	Damp, or too much handling	Replace it. Tape all four edges, taut, and view from outside rather than pressing on it.

Clean-up and what to keep

Untape the blackout, roll it and keep it: Part XVI wants the same material for the darkened corner, and it is annoying to buy twice. Collect the tape ends and card offcuts into one bag. Retract the knife blade and put the knife away in the place you decided on before you started cutting.

Keep the box, and keep it dry and rigid. Write on the outside, in pencil: the internal screen distance in millimetres, the three measured aperture diameters, and the date. Store the three aperture plates inside it in an envelope. Part VI opens with that box and those numbers.

The waste is dry card, tape and a scrap of aluminium foil, carrying nothing that was not already in the packaging it arrived in. Clean foil is accepted by many kerbside collections and tape-contaminated card generally is not, but which streams exist and what may go in them is decided locally and differs between neighbouring authorities. Check your local regulations.

Recap

• Light travels in straight lines, so rays from a scene must cross at a small hole, so the image is turned through 180° — inverted and laterally reversed at once, which is one fact and not two.
• Similar triangles give the image height: h_i = h_o × v / u. Predict before you measure, and treat the gap as information.
• A pinhole has no focus. Moving the screen changes size, angle of view and brightness, never sharpness.
• Geometric blur is about the diameter of the hole; brightness goes as the hole’s area over the square of the screen distance. Both depend on d/v, which is why a room camera and a box camera with the same ratio behave identically.
• Shrinking the hole sharpens the image only until diffraction takes over. That there is a best size is settled; calculating it is Part VI’s job.
• The instrument is at least a millennium old as an eclipse-viewer, is clearly described by Leonardo, is fitted with a lens and a diaphragm in the sixteenth century, and reaches the artist’s portable reflex box by the seventeenth. Every one of those people could see the picture. Keeping it is the rest of this part.

Questions

1. You measure a lamp post 6.0 m high from 24 m away, on a screen 250 mm behind the aperture. What image height do you predict? You measure 58 mm. Give two distinct explanations for the discrepancy, and say what measurement would distinguish between them.
2. A friend proposes doubling the sharpness of your box camera by halving the hole diameter, and restoring the lost brightness by halving the screen distance. Work through what each change does to d/v, to the image size, and to the angle of view, and say what they have actually built.
3. The room camera and the box camera in this build both work at about f/500. State two things that are genuinely different between them, in spite of that, and say which of the two you would choose to photograph a moving subject and why.
4. Explain to someone who has not read this page why the image is upside down and left-right swapped, without using the word “inverted”, and using nothing but the phrase “straight lines”.
5. Ibn al-Haytham wrote that the crescent shape of the eclipsed sun appears only when the hole is very small, and that a very wide hole gives a patch shaped like the hole itself. Explain both halves in terms of this page’s blur argument, and say what the “object” is in each case.
6. This page says that “Vermeer used a camera obscura” and “Barbaro described a diaphragm in 1568” are different kinds of claim. In three sentences, say what kind of evidence would settle each, and why only one of them has it.

Further experiments

• Trace the projection. Tape paper to the wall of the room camera, or over the box’s screen, and draw what you see for twenty minutes. Then write a paragraph on what your eye kept adding that the projection did not contain, and what the projection contained that you left out. Working from an upside-down image is much of the point: it is far easier to draw shapes accurately when your brain has stopped recognising them as objects.
• Project the sun during a partial eclipse. The pinhole projector is the method NASA gives for indirect viewing, and the box camera you have just built is the instrument its own instructions describe. Set it up with the sun behind you, look only at the screen, and read the Hazards section again first. A colander held above a sheet of paper does the same job with a hundred apertures at once and is worth doing alongside.
• The scale test. Build a second box half the depth of the first, pierce its aperture at half the diameter, and compare the two side by side at the same scene. If the argument in this page is right, they should look equally bright and equally sharp for their size. Say what you find, and if they differ, say which term in d/v you would suspect first.
• A night in the room camera. Leave the room camera set up after dark and look at what a street lamp does. Long after the daytime scene has gone, individual bright sources still throw perfectly recognisable images, which tells you something about the difference between resolving a scene and detecting a source.
• The one to wait for. Do not load photographic paper into this box yet. It is light-tight, and the temptation is obvious, but the exposure, the loading and the handling belong to the unfixed photogram and then to Part VI. The camera is finished; the chemistry is not.