Middle School Art & Design

Under the Wave off Kanagawa is about ten inches tall. Katsushika Hokusai designed it around 1830, one sheet in a series called Thirty-six Views of Mount Fuji, and a copy of it sits in the Asian Art collection at the Metropolitan Museum of Art in New York.

Look at what it is made of. The claw of foam at the top is a line. The ribs of the three boats are lines. Mount Fuji, small and quiet in the trough of the wave, is two lines meeting at a point. There is no shading in the way a pencil drawing has shading. Hokusai drew the design, a block cutter carved every one of those lines into cherry wood, and a printer inked the blocks and pressed the paper. Everything that image does to you, it does with edges.

That is worth sitting with, because line is the cheapest thing in art and the one most people stop thinking about first.

What a line actually is

Here is the strange part. Almost nothing in the world has a line around it. Your hand does not have a black outline. The edge of a table is not a stripe of dark; it is the place where one surface stops and another begins.

A line is a mark whose length matters more than its width. When you draw one, you are not copying something that exists. You are inventing a mark that stands for an edge, a boundary, a direction, or a movement. That invention is so useful and so old that every human culture has used it, and every one of us reads it without being taught.

Key idea: A line is not a thing you see in the world. It is a mark that stands for an edge, and everyone already knows how to read it.

Five kinds of line, and what each one says

Lines are not interchangeable. A shaky line and a ruled line describe the same edge and tell you completely different things about it. Here is a working set:

Kind of line	How you make it	What a viewer reads from it
Contour	One continuous line following the outer edge of a thing	Shape, identity, the answer to what is this
Gesture	Fast, loose, several strokes, done in under a minute	Movement, energy, which way a body is leaning
Hatching	Many short parallel lines packed close together	Shadow, weight, a surface turning away from light
Implied line	A row of separate marks, or several people all looking one way	Direction; the eye connects the dots by itself
Ruled or geometric	Straight, even, drawn against a straight edge	Something built, measured or man-made

Test the table against a real thing. Draw a tree trunk with a ruled straight line down each side and it stops reading as a tree; it reads as a post. Draw the same trunk with two slightly wobbling lines that thicken near the ground and it reads as something that grew. You did not add detail. You changed the kind of line.

Why line weight changes the whole drawing

Line weight means how thick and how dark a line is. It is the single easiest thing you can add to a flat drawing to make it stop looking flat, and almost no beginner uses it, because we are all taught to draw with one pencil at one pressure.

Three rules that hold up in almost every case:

• Heavier where a thing sits on something. The line where a mug meets the table should be your darkest line in the drawing. That single dark contact line is what makes an object stop floating.
• Heavier on the shadow side. If light comes from the left, thicken the right-hand contour. Your eye reads the thicker edge as the edge turning away.
• Lighter as things go back. Distant fence posts get thin, pale lines. Near ones get thick, dark lines. That alone creates depth, with no perspective construction at all.

Try it in ninety seconds. Draw a rectangle. Now go back and press hard along the bottom edge and the right edge, and barely touch the top and left. The rectangle lifts off the page. Nothing else changed.

The point: One pencil at one pressure gives you one flat drawing. Varying pressure and thickness is the fastest available upgrade.

Contour drawing: the line that follows an edge

A contour drawing is a drawing made of edges only, with no shading. It sounds primitive and it is brutally hard, because it forces you to look at the object rather than at your memory of the object.

That distinction is the real subject of this lesson. When you draw a hand, most of what your pencil does comes from a stored idea of hand: five fingers, roughly sausage shaped, thumb off to one side. The stored idea is wrong in every particular. Real fingers are wider at the base, the gaps between them are not even, and from most angles you cannot see all five at once. Contour drawing is the exercise that breaks the stored idea.

Blind contour: the exercise that feels like failing

Blind contour drawing is contour drawing without looking at your paper. You look only at the object. Your pencil moves as slowly as your eye moves along the edge, and you do not lift it or peek.

Kimon Nicolaides put this exercise at the front of his 1941 drawing manual and it has been in art classrooms ever since, for one reason: it is the only reliable way to force your eye and your hand onto the same clock. The result always looks wrong. Fingers land in the wrong county. That is expected, and it is not what the exercise is for.

What it trains is the speed of your looking. After ten minutes of blind contour, an ordinary drawing done normally is noticeably better, because you have stopped skimming the object.

Worth holding on to: Blind contour produces bad drawings on purpose. Judge it by what your next normal drawing looks like, not by the blind one.

Hatching: turning line into shadow

Hatching is shading made of parallel lines. Cross-hatching lays a second set across the first at an angle. Pack the lines closer and the area reads darker; spread them out and it lightens.

Two things separate hatching that works from hatching that looks like scribble:

1. Follow the form. On a curved surface, curve the hatch lines the way the surface curves. Hatching a cylinder with straight vertical lines keeps it flat. Hatching it with lines that bow around the barrel makes it round.
2. Change spacing, not pressure, first. Beginners darken by pressing harder, which produces a shiny grey smear. Professionals darken by adding more lines. You keep control of the texture that way.

Look at an etching or a pen drawing from the seventeenth century sometime and you will see this everywhere: every value in the image, from the pale sky to the black under a cloak, made from nothing but ruled quantities of line.

Common misconceptions

• "A good drawing has clean, confident single lines." Some do. Plenty of superb drawings are built from many searching strokes laid over each other, which is how you find an edge rather than guess it. The single confident line usually arrives after the searching, not instead of it.
• "Erasing means I made a mistake." Erasing is a drawing tool. A kneaded eraser pressed onto graphite lifts a highlight out of a shaded area, which is a mark, not a repair.
• "I cannot draw straight lines so I cannot draw." Nobody draws long straight lines freehand well, including professionals, which is why rulers exist. Straightness is a manufacturing skill. Seeing is the drawing skill.

Recap

A line is an invention: a mark standing in for an edge that has no outline in reality. The kind of line you choose, contour or gesture or hatch or ruled, tells the viewer what kind of thing they are looking at before they read a single detail. Line weight, meaning thickness and darkness, is the cheapest way to give a flat drawing depth and contact with the ground. Contour drawing trains you to look at the object rather than your memory of it, blind contour trains that harder by removing your ability to correct, and hatching turns line into value by controlling spacing. Hokusai got a wall of water and a mountain out of carved edges. You have a pencil, which is more forgiving than a block of cherry wood.

Sources

1. The Metropolitan Museum of Art. (n.d.). Under the Wave off Kanagawa (Kanagawa oki nami ura), from Thirty-six Views of Mount Fuji, Katsushika Hokusai, ca. 1830-32. The Met Collection. metmuseum.org
2. Wikipedia contributors. (n.d.). The Great Wave off Kanagawa. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (n.d.). Contour drawing. Wikipedia. en.wikipedia.org
4. Wikipedia contributors. (n.d.). Hatching. Wikipedia. en.wikipedia.org
5. Tate. (n.d.). Art terms glossary. Tate. tate.org.uk
6. Nicolaides, K. (1941). The natural way to draw: A working plan for art study. Houghton Mifflin.

Try this before you read further. Without looking at one, draw a chair from the side. Give it ten seconds.

Almost everyone draws the same wrong chair: a rectangle for the seat, a rectangle for the back, four sticks for legs, all four legs the same length and evenly spaced. It is a symbol of a chair, assembled from the idea of chair. Put a real chair in front of that drawing and the legs disagree, the seat is not a rectangle from any angle, and the back is narrower at the bottom than you drew it.

This lesson is about why that happens and about the single most reliable fix, which is to stop drawing the chair and start drawing the gaps.

Shape and form are not the same word

A shape is flat. It has length and width, it is enclosed by an edge, and that is all. A circle, a triangle, the silhouette of a cat: shapes.

A form is solid. It has length, width and depth, it takes up space, and you could in principle walk around it. A sphere, a cube, a person, a chair: forms.

The painted limestone bust of Nefertiti in the Neues Museum in Berlin, made around 1345 BCE and standing about nineteen inches tall, is a form. You can move to the side and see something you could not see from the front. The photograph of it in a book is a shape. Everything you draw on paper is technically a shape; the whole craft of drawing solid things is the craft of making shapes behave like forms.

What matters here: Shape is flat and bounded by an edge. Form has volume. A drawing is always shapes, arranged so the eye reads form.

Two families of shape

Geometric shapes are the ruled and measured ones: squares, circles, triangles, hexagons. They read as built, designed, human-made. Organic shapes are irregular and curving: a leaf, a puddle, a cloud, the outline of a sleeping cat. They read as grown, natural, alive.

Designers exploit that reading constantly. A hospital logo tends to be geometric because you want it to feel reliable. A juice label tends to be organic because you want it to feel like it came from a plant. Neither family is better; each one carries a message before anyone reads a word.

Negative space, and why your chair went wrong

Now back to the chair. Positive space is the object itself. Negative space is everything else: the triangle of air between two legs, the gap under the seat, the slot between the back slats.

Here is the trap. Your brain has a stored file called chair, and when you draw the object, the file writes the drawing. But your brain has no stored file for the odd trapezoid of air between the front left leg and the back left leg. It has never needed one. So when you draw that trapezoid instead, you are forced to actually look, and the shape you produce is accurate. Then the chair appears around the gaps, correct, almost by accident.

This is not a trick or a shortcut. It is the standard fix for the standard problem, and it works on anything your brain has a symbol for: hands, faces, trees, bicycles, letters.

Why this matters: You cannot draw from a symbol and observe at the same time. Negative space forces observation because there is no symbol to fall back on.

How to draw negative space, in practice

1. Set the object against a plain background so the gaps are obvious. A chair against a blank wall works; a chair in front of a bookshelf does not.
2. Pick one gap, ideally a closed one, such as the hole between the seat, the back and the top of a leg.
3. Draw that gap as if it were the object. Ask how wide it is at its widest and where its corners fall.
4. Move to the next gap and draw it in relation to the first, not to the chair.
5. Only when the gaps are down, darken the edges they share. The chair is already there.

A useful check while you work: compare a gap to a gap. If the triangle of air under the seat is about twice as wide as the slot between the back slats, your drawing should show that ratio. Ratios between gaps are far easier to judge honestly than the length of a leg.

From shape to form: four solids and nothing else

Most objects, for drawing purposes, are built from four solids: the sphere, the cube or box, the cylinder and the cone. Learning to draw those four convincingly gets you most of the way to drawing almost anything, because complicated objects are combinations.

Object	Solids it is built from
A mug	Cylinder, plus a flattened ring for the handle
A human arm	Two tapering cylinders joined at a small sphere
A pine tree	Cone on a narrow cylinder
A car	Two boxes of different heights, four flattened cylinders
A head	A sphere with the sides flattened, plus a box for the jaw

This is why art teachers make you draw so many boring cylinders. It is not devotion to cylinders. It is that once you can turn a cylinder, you can turn an arm.

Three ways to get depth before you learn any perspective

Perspective construction comes later in this course. You do not need it to make a flat page read as deep. Three devices do most of the work:

• Overlap. If shape A covers part of shape B, A is in front. This is the strongest depth cue there is, and it costs nothing. Two circles that overlap are instantly two objects at two distances; two circles that merely sit side by side are two flat shapes.
• Size change. Two things you know are the same size, drawn at different sizes, are read as near and far. Draw three identical people at three sizes and you have built a street.
• Placement. On a page with a ground, things placed higher up read as further away. This is why a horizon works.

Now notice what happens when the devices disagree. Draw a small figure low on the page and a large figure high up, and the picture fights itself: placement says the small one is nearer, size says the large one is. A viewer feels that as wrongness even if they cannot name it. Keeping your depth cues agreeing is most of what people mean by a drawing that reads.

In short: Overlap, size change and placement create depth on their own. They only fail when they contradict each other.

Foreshortening, briefly

Foreshortening is what happens to a form pointed at you: it gets shorter in the drawing than it is in life. An arm reaching toward the viewer may occupy two inches of paper while the arm hanging at the side occupies six. Every beginner draws the reaching arm too long, because the stored file says arms are long.

The cure is measurement, not talent. Hold your pencil at arm's length, close one eye, and mark off how long the reaching arm looks against the pencil. Compare that to the head. If the arm measures shorter than the head, draw it shorter than the head, however wrong it feels.

Common misconceptions

• "Negative space is just the background." Not quite. Background is whatever is behind the subject. Negative space is the specific shape of the empty areas that touch and define the subject, including gaps entirely inside it, like the hole in a mug handle.
• "A form is just a shape with shading added." Shading helps, but overlap, foreshortening and how edges turn do the job as well. Plenty of drawings read as fully solid with no shading at all.
• "If it looks wrong, I need more detail." Almost always the opposite. When a drawing looks wrong, the error is in the big shapes and their proportions, and adding eyelashes to a badly shaped head makes a badly shaped head with eyelashes.

What to carry forward

Shape is flat and enclosed by an edge; form has volume and takes up space. Geometric shapes read as built and organic shapes read as grown, and designers pick between them on purpose. Drawing the negative space, the gaps rather than the object, defeats the stored symbol in your head and is the most reliable accuracy fix available. Most objects can be built from four solids: sphere, cube, cylinder and cone. Overlap, size change and placement give you depth long before you learn any perspective, as long as they agree with one another. And when a form points at you it foreshortens, so you measure rather than trust your sense of how long an arm ought to be.

Sources

1. Tate. (n.d.). Form. Art Terms. tate.org.uk
2. Tate. (n.d.). Foreshortening. Art Terms. tate.org.uk
3. Wikipedia contributors. (n.d.). Negative space. Wikipedia. en.wikipedia.org
4. Wikipedia contributors. (n.d.). Nefertiti Bust. Wikipedia. en.wikipedia.org
5. Tate. (n.d.). Sculpture. Art Terms. tate.org.uk
6. Edwards, B. (2012). Drawing on the right side of the brain (4th ed.). TarcherPerigee.

In 1599 Caravaggio was given a commission for a side chapel in San Luigi dei Francesi, a French church in Rome. He delivered The Calling of Saint Matthew, and it still hangs there, in the Contarelli Chapel, where you can put a coin in a box to light it.

The picture is mostly darkness. A shaft of light comes in high from the right and crosses a bare wall, and out of the black it picks a hand, a face, a feather in a hat, the edge of a shoulder. Caravaggio did not paint a room with people in it. He painted the light, and the room came along.

That is the subject of this lesson. Not colour, not outline. Just how light and dark are distributed, which is the thing that makes a circle turn into a ball.

Value is not colour

Value means how light or dark something is, ignoring what colour it is. A lemon and a cream-coloured wall are different colours with nearly the same value. A navy jumper and a black cat are different colours with nearly the same value.

There is a fast test for this. Photograph anything with a phone and turn the picture to black and white. Whatever still reads clearly in grey was working through value. Whatever collapses into one flat tone was relying on colour alone, and that part of the picture will look weak from across a room, in poor light, or in a photocopy.

This is why almost every art course starts with grey. Get value right and colour becomes decoration you can choose freely. Get value wrong and no amount of beautiful colour rescues the picture.

The core of it: Value is lightness and darkness with colour ignored, and it does most of the work of making things look solid.

Build a value scale before you shade anything

A value scale is a strip of boxes running from the white of the paper to the darkest mark your pencil can make. Five boxes is enough to start with.

1. Draw five boxes in a row, each about one inch square.
2. Leave box 1 as bare paper. That is your white.
3. Fill box 5 as dark as you can, pressing firmly and going over it in two directions until it is even. That is your black.
4. Fill box 3 with something that sits exactly halfway between them. Squint at the strip to judge it; squinting removes detail and leaves value.
5. Fill boxes 2 and 4 with the halfway points between what you already have.

Two things go wrong here for nearly everyone. The first is that box 5 is not dark enough, because pressing hard feels like vandalism. The second is that boxes 2, 3 and 4 end up nearly identical, so the scale has three values pretending to be five. Hold the finished strip at arm's length. If you cannot see five distinct steps from two metres away, it is not a scale yet.

Keep the strip. You will hold it next to your drawings and against real objects for the rest of the course.

What one light does to one ball

Put a plain round object on a table, an orange or a tennis ball, and light it with one lamp from the upper left. One lamp, and turn the other lights off, because two light sources produce two sets of shadows and nothing you can learn from yet.

Now look, and you will find five distinct regions. In order, going away from the lamp:

Region	Where it is	Value on a five-step scale
Highlight	The small bright spot facing the lamp most directly	1, bare paper
Midtone	The broad area sloping away from the lamp	2 to 3
Core shadow	A dark band on the far side, but not at the very edge	4 to 5, the darkest part of the ball
Reflected light	A faint glow at the very outer edge, light bouncing back off the table	3
Cast shadow	The shadow thrown onto the table, darkest where it touches the ball	5 where it touches, lightening as it goes out

The two that people miss are the ones that matter most. The core shadow is not at the outer edge of the ball; it sits slightly inside it, because at the very edge the table is bouncing light back up. And the cast shadow is darkest and hardest right where the ball touches the table, softening and lightening as it stretches away.

Remember: The darkest part of the ball is a band just inside the far edge, not the edge itself. That one fact turns a flat grey disc into a sphere.

Shading the sphere, step by step

1. Draw the circle lightly. Do not outline it darkly; a hard black outline flattens a sphere faster than anything else you can do.
2. Decide where the lamp is and mark the highlight with a small circle. Leave it as bare paper and never shade it.
3. Lay a light, even midtone over everything except the highlight, using hatching that curves around the ball.
4. Find the core shadow band, a crescent set in from the far edge, and build it to value 4 or 5 by adding more layers of hatch, not by pressing harder.
5. Leave or lift a thin strip of reflected light between the core shadow and the outer edge. A kneaded eraser pressed on and lifted works well.
6. Add the cast shadow. Make it darkest at the point of contact and let it fade outward, and note that it stretches away from the lamp.
7. Squint at the whole thing. If the ball and its cast shadow blur into one grey blob, your core shadow and your cast shadow are too similar; darken the point of contact.

The same light, a cube, a different answer

Now change one input. Replace the ball with a cardboard box and light it from the same upper left.

Everything you just learned about smooth gradation is wrong for the cube. A sphere is one continuously curving surface, so its value changes gradually. A cube is three flat planes, and each plane is a single flat value, because every point on a flat plane faces the lamp at the same angle.

So a well-shaded cube is three even tones with hard borders: the top plane light, the side facing the lamp a midtone, the side turned away dark. Blending the corners of a cube destroys it, in exactly the way that failing to blend a sphere destroys that.

Bottom line: Curved surfaces get gradation, flat planes get flat tone. Shading is not a texture you apply; it is a description of which way a surface faces.

Chiaroscuro, and why the dark is doing the work

Chiaroscuro comes from the Italian for light and dark, and it names the use of strong contrast between the two to model solid form. When the contrast is pushed to extremes, with figures emerging from near total blackness, it is often called tenebrism, and Caravaggio is the artist most associated with it.

Go back to The Calling of Saint Matthew. Caravaggio could have lit the whole room evenly. Instead he threw away most of the picture into darkness, which does two jobs at once: it makes the lit hands and faces look genuinely three dimensional, and it tells you where to look. You cannot help but follow the light. Contrast is a pointing device.

Common misconceptions

• "Shading means making things grey." Shading means describing which way each part of a surface faces. If your drawing has one middling grey everywhere, you have described nothing. Real observation almost always turns up a wider range than you expect, from bare paper to genuine black.
• "The shadow goes on the bottom." The shadow goes opposite the light. If the lamp is above and to the left, the shadow is below and to the right. Students who always shade the bottom edge are drawing a memory of shadows, not the shadow in front of them.
• "You have to blend with your finger to make it smooth." Finger blending lifts oils onto the paper and gives a shiny, muddy grey that resists further work. Layered hatching, or a paper stump for small areas, keeps the surface alive.

Putting it together

Value is lightness and darkness with colour set aside, and it is what makes flat shapes read as solid. A five-step value scale, checked from two metres away, gives you something objective to compare against instead of guessing. One light on a sphere produces five readable regions, and the two that beginners miss, the core shadow set in from the edge and the hard dark point where the cast shadow touches, are the two that create roundness. Change the object to a cube and the rule changes with it: flat planes take flat tones. Caravaggio built entire pictures out of that principle, choosing to lose most of the canvas to darkness so the lit parts would stand up and so your eye would go where he sent it.

Sources

1. Tate. (n.d.). Tone. Art Terms. tate.org.uk
2. Tate. (n.d.). Chiaroscuro. Art Terms. tate.org.uk
3. Wikipedia contributors. (n.d.). The Calling of Saint Matthew (Caravaggio). Wikipedia. en.wikipedia.org
4. Wikipedia contributors. (n.d.). Chiaroscuro. Wikipedia. en.wikipedia.org
5. Wikipedia contributors. (n.d.). Shading. Wikipedia. en.wikipedia.org

In 1824 a French industrial society offered six thousand francs to anyone who could manufacture ultramarine blue artificially. Four years later Jean-Baptiste Guimet collected it. Before that, the only source of true ultramarine was lapis lazuli, a stone mined mostly in what is now Afghanistan, ground and purified at enormous cost. Painters used it for the Virgin's robe and for very little else, because a patron had to pay for it separately, by weight.

Today a tube of ultramarine costs less than a sandwich. The colour problem has moved. It is no longer about getting hold of a pigment; it is about what happens when you put two of them together and get something horrible.

Why your purple turned brown

Nearly every student has this experience. You want purple. You have red and blue. You mix them, and out comes a dull brownish mauve that looks nothing like the purple in your head.

Nothing is wrong with you or the paint. The explanation is that the red in a basic set is usually a warm red leaning toward orange, which means it already contains some yellow. Add blue, and you have quietly mixed all three primaries. All three primaries together is the recipe for brown. Your purple was contaminated with yellow before you started.

The fix is not more paint. It is a different red. A cool red leaning toward pink, sometimes sold as crimson or magenta, contains no yellow, and mixed with blue it gives a real purple. Same technique, different starting colour, completely different result.

So what?: Most bad mixes are caused by what your paints already contain, not by how you mixed them.

Three words that fix most colour problems

Colour is not one property. It is three, and separating them turns a vague complaint into a fixable problem.

Property	What it means	How you change it
Hue	The name of the colour: red, orange, blue-green	Mix in a neighbouring hue on the wheel
Value	How light or dark it is	Add white to lighten, add a dark colour to darken
Saturation	How intense or how dull it is	Add a little of the colour opposite it on the wheel

Now the diagnosis becomes easy. If your sky is the wrong colour, ask which of the three is off. Is it the wrong hue, too green rather than too blue? Is it the right hue but too pale, a value problem? Or is it the right hue and right lightness but screaming, when the real sky is quieter, which is a saturation problem? Each answer sends you to a different jar.

Most beginners only ever adjust one of the three. They lighten and darken, endlessly, when the real fault is that everything is at full saturation, which is why student paintings often look loud.

The wheel, and what it is a map of

The colour wheel arranges hues in a circle so that colours which mix well sit near each other and colours which cancel each other sit opposite. In the traditional painter's version:

• Primary colours are red, yellow and blue. You cannot mix them from anything else in this system.
• Secondary colours come from mixing two primaries: orange, green and purple.
• Tertiary colours come from mixing a primary with its neighbouring secondary, giving red-orange, yellow-orange, yellow-green, blue-green, blue-purple and red-purple.
• Complementary colours sit directly opposite each other: red and green, blue and orange, yellow and purple.

An honest note, because you will meet this later and it confuses people. The red-yellow-blue wheel is a workshop tool, not a law of physics. Printers use cyan, magenta and yellow, which mix a wider range of colours than red, yellow and blue do; screens use red, green and blue light. All three systems are real and each is right for its own material. Red, yellow and blue is what most paint sets are built around, so it is what we will use.

A worked mix, one colour at a time

Say you want the green of a summer leaf, which is a slightly yellowish green, medium in value, and not fully saturated.

1. Start with the hue. Mix yellow and blue. Begin with more yellow than blue, because blue overpowers yellow, and a common error is to dump in blue and then need half the tube of yellow to climb back out.
2. Check the hue against a real leaf. Too blue and cold? Add yellow, a little at a time. Too acid and yellow? Add the tiniest touch of blue.
3. Now the value. Hold it next to the leaf and squint. If your green is darker than the leaf, add a little white; if lighter, add a touch of blue or a dark colour rather than black.
4. Finally saturation. Fresh mixed green is usually too intense for a real leaf. Add a speck of red, the complement of green, and it settles down without going grey.

Do those in order. If you adjust saturation before hue you will chase your own tail, because dulling a wrong hue just gives you a dull wrong hue.

Key idea: Fix hue first, then value, then saturation. Out of order, each change undoes the last one.

Dulling a colour without killing it

There are two ways to make a colour less intense and they do not give the same result.

Adding black lowers value and saturation at once, and it tends to deaden a colour, turning bright red into a flat maroon and yellow into a strange olive. Adding a small amount of the complement lowers saturation while keeping the colour alive, because the mixture still contains the original pigment plus a competing one, and the eye reads that as a colour rather than as dirt.

Push it further and the complementary pair produces a warm or cool grey, sometimes called a chromatic grey, which is far more interesting than grey from a tube. Painters use this constantly for shadows: a shadow on a red apple is often the red mixed toward green, not the red mixed toward black.

Why paint and light behave in opposite directions

Mix red paint and green paint and you get brown. Shine a red light and a green light on the same white wall and you get yellow. Both are correct, and the reason is worth knowing.

Paint is subtractive. A red pigment looks red because it absorbs most of the light that hits it and reflects back the red. Add a second pigment and you are adding a second absorber, so between them they subtract more of the light and less comes back. Keep adding and you approach black, which is why mixing four or five paints together nearly always gives mud.

Light is additive. Adding a second beam adds energy, so the result is brighter, and red, green and blue light together make white. This is what your screen does with three tiny sub-pixels at every point.

The upshot: Paint gets darker and duller as you add more; light gets brighter. Keep mixes to two or three colours plus white.

Common misconceptions

• "Black and white are colours you mix with freely." White is genuinely useful, though it lowers saturation as it lightens. Black is a trap: it flattens most colours. Try mixing your darks from a dark blue plus a brown or a red before reaching for black.
• "There are exactly three primary colours." There are three in whichever system you are using, and the systems disagree. Paint and printing and screens use different sets because pigment and light behave differently.
• "A colour looks the same wherever you put it." It does not. The same grey square looks lighter on a black background and darker on a white one, and a colour placed next to its complement looks more intense than the same colour placed next to a neighbour. Josef Albers built a whole teaching method on that effect.

The short version

Colour has three separable properties, and naming which one is wrong is most of the work: hue is which colour, value is how light, saturation is how intense. The wheel maps hues so that mixing partners sit near each other and cancelling partners sit opposite. Fix hue, then value, then saturation, in that order. Dull a colour with a touch of its complement rather than with black, and use complementary mixtures for shadows. Paint is subtractive, so it darkens and dulls as you add more, which is why two or three colours plus white is a working limit. And when a mix fails, suspect your starting pigments first, the way a warm red with yellow hiding in it ruins every purple you try to make from it.

Sources

1. Wikipedia contributors. (n.d.). Color wheel. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). Primary color. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (n.d.). Subtractive color. Wikipedia. en.wikipedia.org
4. Wikipedia contributors. (n.d.). Ultramarine. Wikipedia. en.wikipedia.org
5. Tate. (n.d.). Complementary colours. Art Terms. tate.org.uk
6. Albers, J. (1963). Interaction of color. Yale University Press.

Vincent van Gogh painted Wheat Field with Cypresses in 1889, and one version of it hangs in the European Paintings galleries of the Metropolitan Museum of Art. It is a hot gold field under a churning blue-white sky, with a dark green cypress standing up the right-hand side like a flame.

Count the colours. There are essentially two families in the picture, gold and blue, sitting opposite each other on the wheel, plus the green of the tree holding them apart. That is not accidental and it is not complicated. It is a scheme, chosen, and it is doing more for the painting than any amount of detail could.

Warm, cool, and what temperature does to space

Split the wheel in half. Reds, oranges and yellows are called warm. Blues, greens and purples are called cool. The names come from association, fire and sun on one side, water and shade on the other, and they are useful for a reason that has nothing to do with temperature.

Warm colours tend to advance and cool colours tend to recede. Put an orange square and a blue square of the same size and value side by side on white paper, step back, and the orange will look slightly nearer. Landscape painters have used this for centuries: distant hills are painted cooler and bluer than nearby ones, partly because the air really does scatter blue light over distance, and partly because a cool blue hill sits back in the picture whether or not the physics is on display.

Temperature is also relative. There is no absolute line. A red can be warm next to a blue and distinctly cool next to an orange, in the same way that a lukewarm drink is cold after a hot one.

What matters here: Warm advances, cool recedes, and both judgements depend on what is next to them.

Five schemes, and what each is for

A colour scheme is a rule you set yourself about which part of the wheel you will use. The rule is the point: limiting your choices is what makes a picture look designed rather than assembled.

Scheme	The rule	Feels like	Use it when
Monochromatic	One hue, varied in value and saturation only	Calm, unified, sometimes flat	You want mood over drama, or you are practising value
Analogous	Three neighbours on the wheel, such as yellow, yellow-green, green	Harmonious, natural, easy on the eye	Landscapes, backgrounds, anything that should not shout
Complementary	Two opposites, such as blue and orange	High energy, strong contrast	You need one thing to leap off the page
Split complementary	One hue plus the two neighbours of its opposite	Nearly as punchy, less jarring	You want contrast without a head-on clash
Triadic	Three hues evenly spaced around the wheel	Lively, balanced, playful	Posters, packaging, anything aimed at children

Now read the table against the Van Gogh. Gold and blue are opposites, so it is a complementary scheme, and the table predicts high energy and strong contrast, which is exactly what that painting delivers. Read it against a photograph of a forest, all yellow-greens and greens and blue-greens, and you get an analogous scheme, which the table predicts will feel harmonious and calm. The table is not a set of laws. It is a set of consequences you can plan for.

Dominance: why equal amounts fail

Here is the single most common scheme failure. A student picks blue and orange, then uses roughly half of each. The result is restless and slightly unpleasant, and they conclude the scheme was wrong.

The scheme was fine. The proportions were wrong. Complementary colours in equal amounts fight for the same job, which is to be the thing you look at, and neither wins. A useful habit borrowed from interior design is roughly sixty, thirty, ten: about sixty percent of the area in a dominant colour, thirty in a supporting one, and ten in an accent. Under that rule, blue and orange becomes a mostly blue picture with orange in the places you want the eye to land.

Look back at the Van Gogh with the proportions in mind. The gold field and the pale sky take most of the canvas. The dark green cypress is a small fraction of it. The small area is the one you look at first.

Worth holding on to: A scheme is a set of colours plus a set of proportions. Equal amounts of two complements cancel each other out.

The same colour, twice, looking different

Cut two identical small grey squares from the same sheet. Put one on a black background and one on a white background, and step back. The one on black looks lighter. Nothing about the grey has changed.

This effect is called simultaneous contrast, and Josef Albers built an entire teaching method around it at Yale, published as Interaction of Color in 1963. His central claim was blunt: colour is the most relative medium in art, and you cannot know what a colour will do until you see it in position.

The practical consequences are immediate. Judging a paint mix on a white palette tells you very little, because it is surrounded by white. Test a mix by putting a dab of it next to the colours it will actually sit beside. And if a colour in your painting looks wrong, consider changing its neighbour instead of changing it.

Colour and mood, with the caveats

Colour clearly affects how a picture feels. What is far less clear is that particular colours carry fixed meanings.

Some associations are widely shared within a culture: in much of Europe and the Americas, red signals danger or urgency on signs and warnings, and blue is used constantly by banks and hospitals to suggest steadiness. But the associations are not universal. White is the standard colour of Western weddings and is a traditional colour of mourning in parts of East Asia. Purple has meant royalty in some periods because the dye was ruinously expensive, which is an economic fact rather than a psychological one.

So the honest version is this. You can rely on temperature and contrast, which are about how eyes work. Be more careful with claims like green means envy, which are about a particular culture at a particular time. If you are designing something for a specific audience, the useful question is not what does this colour mean but what does this colour mean to them.

Common misconceptions

• "A complementary scheme means using lots of both colours." It usually means one colour holding most of the area and its opposite used sparingly, where you want attention. The contrast is a tool for pointing, and it stops pointing if it is everywhere.
• "Colours have fixed emotional meanings." Some associations are strong within a culture and some are learned from a specific industry, but very few are universal, and designers who assume otherwise get surprised.
• "If a colour looks wrong, change that colour." Often the neighbour is the problem. Simultaneous contrast means a colour is partly created by what surrounds it, so shifting the background can fix a colour that is already correct.

Where this leaves us

Warm colours advance and cool colours recede, which gives you depth control before you touch perspective, and both judgements are relative to whatever sits nearby. A colour scheme is a self-imposed rule about which slice of the wheel you will use, and the five standard ones each have predictable consequences: monochromatic is calm, analogous is harmonious, complementary is loud, split complementary softens that clash, triadic is lively. Proportion matters as much as the palette, and roughly sixty, thirty and ten is a reliable starting split. Colour is relative, as Albers spent a career demonstrating, so test mixes in position rather than on a white palette. And when you reach for the emotional meaning of a colour, remember that most of those meanings were taught rather than born.

Sources

1. The Metropolitan Museum of Art. (n.d.). Wheat Field with Cypresses, Vincent van Gogh, 1889. The Met Collection. metmuseum.org
2. Wikipedia contributors. (n.d.). Josef Albers. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (n.d.). Contrast effect. Wikipedia. en.wikipedia.org
4. Tate. (n.d.). Complementary colours. Art Terms. tate.org.uk
5. Wikipedia contributors. (n.d.). The Starry Night. Wikipedia. en.wikipedia.org
6. Albers, J. (1963). Interaction of color. Yale University Press.

In 1925 Max Ernst was staying in a seaside inn and found himself staring at the floorboards, which had been scrubbed so often that the grain stood up in ridges. He put a sheet of paper on them and rubbed a soft pencil across it. The grain came through as a pattern of dark and light, and Ernst started seeing things in it: forests, birds, heads.

He called the technique frottage, from the French for rubbing, and made a whole series of works from it. What matters for you is cheaper than that. Ernst had discovered that a surface can be captured and moved, and that a texture on paper carries the memory of the thing it came from.

Actual texture and implied texture

Actual texture is real. It is how a surface feels under your fingers: the ridge of thick paint, the tooth of watercolour paper, the roughness of a sculpture in unpolished stone. If you could touch it with your eyes shut and tell the difference, it is actual.

Implied texture is a picture of a texture. It feels perfectly smooth. A pencil drawing of a sheep is flat paper, but the marks say wool. A photograph of rusted metal is glossy and even to the touch and it says rust anyway.

Most drawing is implied texture, and getting good at it is a matter of choosing marks rather than adding effort. This is the part that surprises people. Drawing a convincing sheep is not about drawing more wool.

Key idea: Actual texture you can feel. Implied texture is marks that make a flat surface read as a rough one.

Collecting textures with a rubbing

A rubbing is the cheapest texture tool there is, and it takes about fifteen seconds.

1. Find a surface with a raised or sunken pattern: a coin, a leaf with strong veins, a brick, the sole of a shoe, a wire mesh, a paving stone.
2. Lay thin paper over it. Copy paper works; tracing paper works better.
3. Hold the paper still with one hand. With the other, rub the side of a pencil lead, not the point, evenly across the whole area.
4. Rub in one direction first, then lightly in a second direction to fill gaps.

The two mistakes are pressing so hard that the paper tears on the ridges, and using the pencil point, which gives you scratchy lines instead of an even lift. Use the side of the lead and a light, quick motion.

Collect eight or ten of these and you have a texture library you can cut up and paste into collages, or study to work out what marks a texture is actually made of.

How to draw a texture without drawing all of it

Here is the principle that separates a convincing drawn texture from a tedious one: suggest at the edges, and let the middle go quiet.

Look at how a good drawing of a wool jumper works. The artist draws the fuzzy, broken outline carefully at the edges of the sleeve, adds a handful of wool marks in the shadowed folds where the texture would catch the light, and leaves large areas of the middle almost blank. Your eye reads the whole jumper as woolly, because you saw wool at the edges and had no reason to doubt it.

Draw every fibre and two bad things happen. It takes four hours, and the texture becomes so even that it flattens: an area of uniform detail reads as a grey patch, exactly like an area of uniform shading.

The point: Texture is convincing where the eye checks, which is at edges and in shadows. Everywhere else you can leave it alone.

Four surfaces and the marks that make them

Surface	The mark that says it	The mistake to avoid
Fur or wool	Short broken strokes that follow the direction it grows, and a fuzzy silhouette	Drawing each hair evenly across the whole animal
Brick	A few full bricks near the viewer, then just the horizontal mortar lines further off	Ruling every brick on the whole wall
Glass or metal	Hard-edged patches of near-white next to near-black, with almost nothing in between	Smooth grey shading, which reads as plastic
Bark or stone	Irregular vertical cracks with dark inside them, thicker on the shadow side	Even scribble, which reads as noise

Notice that two of these rows are about contrast rather than pattern. Glass and polished metal have almost no texture at all; what makes them look like themselves is the jump from very light to very dark with no middle ground. If you shade a spoon smoothly it turns into grey plastic, however carefully you draw the shape.

Texture is a distance cue

Stand at one end of a gravel path. At your feet you can see individual stones. Ten metres away the stones merge into a speckle. Fifty metres away the path is a smooth grey band.

That fall-off is called a texture gradient, and your eye uses it to judge distance without being asked. Which means you can use it deliberately. Detailed texture in the foreground, simplified in the middle, none at the back, and the picture acquires depth even if you have drawn no perspective lines at all. Reverse it, putting sharp detail on distant objects and none nearby, and the picture goes strange in a way viewers can feel but rarely name.

Texture you can actually feel

When you leave the pencil behind, texture stops being implied and becomes real.

• Collage means gluing found materials onto a surface: newspaper, fabric, sandpaper, foil. The materials keep their own actual texture, which is exactly why artists reached for them.
• Impasto means paint applied thickly enough to stand up off the surface and hold the mark of the brush or knife. Look at a Van Gogh in a gallery from the side, rather than straight on, and the paint has physical height.
• Mixed media just means a work using more than one material, which is the usual case once you start adding sand to paint or stitching thread through paper.

Bottom line: If a texture matters enough, you can stop drawing it and start attaching it.

Common misconceptions

• "More detail makes a texture more convincing." Past a point it does the opposite. Uniform detail across a large area averages out into flat grey, so the wool stops looking like wool and starts looking like tone.
• "Smooth objects have no texture to draw." Smooth objects have the most demanding texture of all, because their surface is described entirely by the sharpness of the value jumps. Soften those and glass becomes plastic.
• "Rubbings are just for children." Ernst built a body of work on them in the 1920s, and rubbings remain a standard way to record raised surfaces, from gravestone inscriptions to the pattern on a manhole cover.

What to remember

Texture comes in two kinds: actual, which you can feel, and implied, which is marks on a flat surface doing an impression of feeling. Rubbings capture an actual texture and turn it into raw material in seconds, using the side of the lead rather than the point. Drawn texture works best when it is concentrated where the eye checks, at edges and in shadows, and left alone in the middle, because uniform detail flattens into grey. Different surfaces need different marks, and the shiny ones are described by hard jumps between light and dark rather than by pattern. Texture also falls off with distance, so grading it from detailed in front to plain at the back builds depth for free. And when implied texture is not enough, collage and thick paint give you the real thing.

Sources

1. Wikipedia contributors. (n.d.). Rubbing (art). Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). Max Ernst. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (n.d.). Texture (visual arts). Wikipedia. en.wikipedia.org
4. Tate. (n.d.). Collage. Art Terms. tate.org.uk
5. Tate. (n.d.). Mixed media. Art Terms. tate.org.uk

Around 1427, Masaccio painted a Holy Trinity on a wall of the church of Santa Maria Novella in Florence. Visitors reported that the wall appeared to have a chapel cut into it: a barrel-vaulted room receding backwards into solid stone, complete with coffered ceiling and columns.

There was no chapel. There was a flat wall, some plaster, and a geometric method that had been worked out in Florence in the previous few years, in large part by the architect Filippo Brunelleschi. Leon Battista Alberti wrote the method down in 1435 in a treatise called On Painting and dedicated it to Brunelleschi, and after that anyone who could read could build a convincing room out of straight lines.

This lesson is that method, reduced to what you need. It is not difficult. It is fussy, which is different.

Two things before you draw anything else

Every perspective drawing rests on two decisions, and both are made before any object appears.

The horizon line is a horizontal line across your page representing your eye level. Not the middle of the paper. Not where the ground meets the sky in a photograph you half remember. It is the height your eyes are at, extended out into the scene forever.

Prove it to yourself. Sit down and look at a table. You can see the top surface, so your eyes are above the table, and the table sits below your horizon line. Now crouch until your eyes are level with the tabletop. The top vanishes; the table becomes a line. Lie on the floor and you see the underside. Nothing moved except your eye level.

The vanishing point is a single dot on the horizon line where all the lines that run directly away from you appear to meet. It sits directly in front of your eyes. In one-point perspective there is exactly one of them.

Why this matters: Where you put the horizon line decides whether the viewer is a small child, a standing adult or a bird. Move it and the whole drawing changes meaning.

The picture, in one diagram

Four lines run from the corners of the back wall out to the corners of the page. Those are the walls, floor and ceiling of the corridor, and every one of them, extended backwards, meets at the vanishing point.

Building the corridor, step by step

1. Draw a horizontal line across the page, roughly a third of the way down if you want the viewer standing, or two thirds down if you want them sitting. This is the horizon.
2. Mark a dot on it. That is the vanishing point. Put it off-centre; dead centre gives a stiff, symmetrical picture.
3. Draw a rectangle somewhere around the vanishing point. This is the far end of the corridor, the back wall.
4. With a ruler, draw a line from each corner of that rectangle through the vanishing point and onwards to the edge of the paper. You now have the corridor. The four wedges are the two side walls, the floor and the ceiling.
5. Add a door on the left wall. The top and bottom edges of that door must run along lines that head to the vanishing point. Its two vertical edges stay exactly vertical.
6. Add a window on the right wall the same way.
7. Erase your construction lines where they cross the walls, and darken the lines you are keeping.

That is a room. Six steps, one ruler.

The three families of line, and the rule that governs them

Family	Example in the corridor	Rule
Lines running away from you	The join between wall and floor, the top edge of a door	Go to the vanishing point
Horizontal lines across your view	The bottom of the back wall, a picture frame on the back wall	Stay horizontal
Vertical lines	The corner where two walls meet, the sides of a door	Stay vertical

Almost every mistake in a beginner's perspective drawing is a line put in the wrong family. If you find yourself angling a door's vertical edge, stop: verticals stay vertical in one-point perspective, always.

Spacing things as they go back

Here is where most drawings quietly fail. You draw floor tiles receding to the vanishing point, then space the crosswise lines evenly, and the floor looks like a ladder lying down. Real receding spaces get smaller, and they do it at a particular rate.

There is a construction that gets it right and needs no measuring:

1. Draw the first tile as a rectangle at the front of the floor, with its side edges running to the vanishing point.
2. Draw both diagonals of that rectangle. Where they cross is its true centre, in perspective.
3. Draw a horizontal line through that crossing point. That is the halfway line of the tile, correctly foreshortened.
4. To add the next tile going back, draw a line from the front bottom corner of the tile through the midpoint of the far edge, and continue it until it hits the receding side line. That intersection gives you the far edge of the next tile.
5. Repeat. Each new tile is shallower than the last, automatically, in exactly the right ratio.

The same construction spaces fence posts, telegraph poles, ceiling beams and the pillars in a cathedral. It is the difference between a drawing that reads as a corridor and one that reads as a diagram of a corridor.

Remember: Receding spaces shrink. The diagonal construction gives you the correct rate without a single measurement.

Putting people in the room

There is a rule here that feels like a magic trick the first time you use it.

If everyone in your picture is roughly the same height as the viewer, and everyone is standing on the same flat floor, then all of their eyes sit on the horizon line. All of them. The near figure is drawn tall and the far figure is drawn small, but the tops of their heads land at the same height, because their eyes are all at your eye level.

So to place a figure correctly: decide where their feet touch the floor, then draw upwards until their eyes reach the horizon. That fixes their size. A figure whose eyes float above the horizon is a giant or is standing on something. A figure whose eyes are below it is a child, or kneeling, or your drawing is wrong.

Common misconceptions

• "The horizon line is where the ground meets the sky." Only when you are outdoors on flat ground. Indoors it is still there, running invisibly through the room at your eye level. In a picture of a mountain valley it may be hidden behind a mountain and still governs every line.
• "Everything in the picture goes to the vanishing point." Only the lines running directly away from you. Horizontals stay horizontal and verticals stay vertical, and drawings where every line converges look like they are being sucked down a drain.
• "Perspective is a rule about how things really look." It is one method, developed in fifteenth-century Italy, for representing a scene from a single fixed eyepoint. Japanese prints of the same era use a different and internally consistent system, and neither is a mistake.

Pulling it together

One-point perspective needs two decisions before it needs any drawing: a horizon line at the viewer's eye level, and a vanishing point on it. From those, four ruled lines through the corners of a back wall produce a room. Lines running away from the viewer meet at the vanishing point; horizontals stay horizontal; verticals stay vertical, and mixing up those three families is where almost all beginner errors live. Spaces get shallower as they recede at a rate you can construct with diagonals rather than guess. Figures standing on the same floor share a single eye height, the horizon, however far away they are. Masaccio used all of this on a church wall in Florence six hundred years ago and made people think a chapel had appeared.

Sources

1. Wikipedia contributors. (n.d.). Perspective (graphical). Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). Holy Trinity (Masaccio). Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (n.d.). Filippo Brunelleschi. Wikipedia. en.wikipedia.org
4. Wikipedia contributors. (n.d.). De pictura. Wikipedia. en.wikipedia.org
5. Tate. (n.d.). Perspective. Art Terms. tate.org.uk

The Baptistery of San Giovanni in Florence is an octagon. Because it has eight sides, there is almost nowhere you can stand and see only one of them; from the cathedral steps opposite, you see two faces meeting at a corner, one running off to the left and one to the right.

According to his biographer Antonio Manetti, Brunelleschi painted that exact view on a small panel around 1425, drilled a hole through it, and had people look through the hole from the back while holding a mirror in front. Take the mirror away, and the real Baptistery was there. Put it back, and so was the painting. He was demonstrating that his construction produced the same image the eye did.

What he was looking at is a two-point problem, and it is the one you will actually need, because most of the world does not politely face you square on.

When one point stops being enough

One-point perspective works when one face of your subject is parallel to the picture plane, meaning it faces you square on: a corridor you are standing in the middle of, a row of shop fronts photographed straight across the street.

Turn forty degrees and everything changes. Now no face is parallel to you. The left wall runs away to the left and the right wall runs away to the right, and each needs its own vanishing point. That is two-point perspective.

Both vanishing points sit on the same horizon line, because both walls stand on the same flat ground and rise at the same angle. This is the single fact that keeps a two-point drawing from falling apart.

What matters here: Two-point means two vanishing points, and both live on the one horizon line.

The setup, in one diagram

One thick vertical in the middle is the corner of the building nearest you. From its top and bottom, two lines run left toward the left vanishing point and two run right toward the right one. Where you stop them decides how wide each wall is.

Drawing the building, step by step

1. Rule a horizon line across the page. Put it about a third of the way down for a normal standing view.
2. Mark two vanishing points on it, one near each edge of the paper, or better still just off the paper on scrap taped to either side.
3. Draw one vertical line between the vanishing points, somewhere in the middle third of the page. This is the corner of the building closest to you. Its height sets the scale of everything.
4. From the top of that vertical, rule a light line to the left vanishing point and another to the right one. Do the same from the bottom. You now have two wedges.
5. Decide how far each wall extends and drop a vertical inside each wedge to close it. Those two verticals are the far corners of the building. They must be vertical.
6. Add windows and doors. Every horizontal edge on the left wall runs to the left vanishing point; every horizontal edge on the right wall runs to the right one. All vertical edges stay vertical.
7. For a pitched roof, find the centre of the end wall by crossing its diagonals, draw a vertical up from that crossing point, and put the ridge on it.
8. Erase construction lines and vary your line weight, darkest on the near corner.

Notice which lines are furthest from the vanishing point: the near corner is the tallest vertical in the drawing, and each wall shrinks as it runs away. If your far corner is taller than your near corner, something has gone wrong at step 5.

Change one input, and watch it go wrong

Now redraw the same box with the two vanishing points close together, say only a hand's width apart. The result looks broken. Corners come out sharply pointed, the box looks stretched and bent, and it reads like a photograph taken through a fisheye lens.

The reason is the cone of vision. Your eyes take in roughly sixty degrees of comfortable, undistorted view. Close vanishing points correspond to a very wide angle of view, wider than that comfortable cone, and everything near the edges of such a view stretches.

So the practical rule is the opposite of what feels convenient: push the vanishing points as far apart as you can, ideally off the paper. Tape spare sheets to the left and right of your drawing and put the points on those. A box drawn with vanishing points a metre apart looks calm and normal. The same box with points twenty centimetres apart looks like it is melting.

In short: Distortion in perspective is almost always vanishing points placed too close together.

Moving the horizon, and what the viewer becomes

Horizon position	What you see	How it feels
High on the page	The tops of things; you look down on the building	Detached, map-like, a bird's eye view
About a third down	Normal standing view of a building on a street	Ordinary, everyday
Very low, near the bottom	The undersides of things; the building towers	Dramatic, imposing, a worm's eye view

Comic artists and film directors move eye level constantly for exactly this reason. Drop the camera to knee height and an ordinary person becomes a looming figure; raise it above their head and they become small and vulnerable. You have the same control, and it costs you one decision at the start of the drawing.

Drawing a street rather than a box

A street is not harder than a box; it is the same box repeated. The key fact is this: any building that is lined up parallel to the first one shares its two vanishing points. A whole terrace of houses along a straight road uses one pair of points for the lot.

A building set at a different angle, the pub on the corner turned to face the junction, gets its own pair of vanishing points, on the same horizon line but in different places. That is why a drawing of a real street may have four or six vanishing points scattered along one horizon.

To draw a receding row of identical windows or lamp posts, use the diagonal method from the previous lesson: it works exactly the same way on a wall running to a vanishing point as it did on a floor.

Common misconceptions

• "Two-point perspective has two horizon lines." There is one horizon line, always, because there is one eye level. Both vanishing points sit on it. Two horizons means two viewers.
• "More vanishing points is more realistic." Three-point perspective exists and is used when you tilt your head up at a skyscraper, adding a third point above or below for the verticals. But for ordinary standing views, verticals really do stay parallel, and forcing them to converge exaggerates the drawing.
• "Perspective must be done freehand to look artistic." Ruled construction lines, erased later, are the normal professional method. Architects, animators and comic artists build with rulers and then draw over the top.

What you now know

Two-point perspective is what you need when you are looking at a corner rather than a face, and it uses two vanishing points sitting on a single horizon line at your eye level. The drawing starts from one vertical, the nearest corner, and grows outwards along lines aimed at each point, with all verticals staying strictly vertical. Placing those vanishing points close together produces the melted, fisheye look, so push them apart and off the page. Where you put the horizon decides whether the viewer looks down on the scene or up at it, which is a storytelling decision as much as a technical one. Buildings parallel to each other share vanishing points; buildings turned at a different angle get their own. And Brunelleschi checked all of this six hundred years ago with a painted panel, a drilled hole and a mirror.

Sources

1. Wikipedia contributors. (n.d.). Vanishing point. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). Filippo Brunelleschi. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (n.d.). Perspective (graphical). Wikipedia. en.wikipedia.org
4. Wikipedia contributors. (n.d.). Isometric projection. Wikipedia. en.wikipedia.org
5. Tate. (n.d.). Perspective. Art Terms. tate.org.uk

In 1797 an English engraver named John Thomas Smith published a book called Remarks on Rural Scenery. In it he argued that a picture divided by thirds, rather than in half, produced a more agreeable balance of light and dark, and the phrase rule of thirds entered the language from that page.

It has since been printed on the viewfinder of almost every camera ever sold, taught in every beginner photography class, and repeated so often that it has stopped sounding like one person's opinion from 1797 and started sounding like physics. It is not physics. It is a good default with well-known exceptions, and knowing which is which is what this lesson is for.

What composition is actually deciding

Composition is the arrangement of things within the frame. Every composition decision answers one of three questions: where does the viewer's eye go first, where does it go next, and does it stay in the picture or fall out of the edge.

That last one is worth pausing on. A picture can leak. If a strong diagonal runs off the bottom right corner and nothing brings the eye back, viewers follow it out and are done with your picture in under a second. Composition is partly a system of walls.

Remember: Composition controls the order in which things are seen, and whether the eye stays inside the frame.

The rule of thirds, stated properly

Divide the frame into nine equal boxes with two evenly spaced vertical lines and two horizontal ones. The rule says: place important elements along those lines, and place the single most important element at one of the four points where the lines cross.

Two practical versions you will actually use:

• Landscape. Put the horizon on the upper third line or the lower third line, not across the middle. Low horizon gives you a picture about sky. High horizon gives you a picture about ground.
• Portrait. Put the subject's eyes on the upper third line, and if the person is looking to one side, leave the larger empty space on the side they are facing.

Why does it help? Mostly because it prevents the two commonest weaknesses at once. Dead centre splits a picture into two equal halves that compete, and it removes any sense of direction. And an object stuck exactly in the middle stops the eye immediately rather than moving it around the frame. A third-point placement gives you asymmetry, which reads as movement, and it leaves room in the frame for the subject to relate to something else.

The case for centring, which is not weak

Now the other side, with actual evidence.

Leonardo's Last Supper, painted in the refectory of Santa Maria delle Grazie in Milan in the 1490s, puts Christ dead centre, with the vanishing point of the whole architectural construction landing on his head. The symmetry is the meaning: the still centre with chaos radiating outward on both sides. Raphael's School of Athens, painted in the Vatican between 1509 and 1511, does the same thing, framing Plato and Aristotle in the exact middle of a receding arch.

Neither of these is an accident or an artist who had not heard of thirds, since neither had, Smith being three centuries later. They are pictures where symmetry does a job: stability, ceremony, authority, stillness. Photographers use it for the same reasons, which is why a portrait meant to feel formal and confrontational is often dead centre and a portrait meant to feel candid is not.

Bottom line: The rule of thirds is a reliable default that prevents common errors. Symmetry is a deliberate choice that produces stillness and authority. Neither is a law.

Five tools that are not the rule of thirds

Tool	What it does	How to use it
Leading lines	Guides the eye toward the subject	A road, a fence, a shadow, a row of windows aimed at what matters
Framing	Isolates the subject and adds depth	Shoot or draw through a doorway, an arch, overhanging branches
Contrast as focal point	Makes one thing the answer to where do I look	The lightest area, or the only saturated colour, or the sharpest edge
Repetition and rhythm	Creates pattern, and makes the break in it powerful	Seven identical arches and one that is different; the eye goes to the odd one
Asymmetrical balance	Balances without mirroring	One large quiet shape on the left against a small intense shape on the right

That last row deserves a moment. Balance does not mean equal amounts on both sides. Imagine a seesaw: a heavy weight close to the middle balances a light weight far out. In a picture, a large area of dull colour balances a small area of bright colour, and a big soft shape balances a small sharp one. This is how most good compositions are balanced, and it is far more useful than mirroring.

Cropping: the revision that costs nothing

Nearly every weak composition improves when something is removed. Cropping is choosing a new frame inside the one you have, and it is the fastest fix available.

Make yourself two L-shaped pieces of card. Overlap them over a drawing or a photograph and slide them around to try out different frames. You will discover, repeatedly, that the interesting picture was hiding inside the boring one, usually with about a third of the original cut away.

Three specific things to look for while you slide the cards:

1. Objects growing out of heads, poles that appear to sprout from someone's skull. Move the frame slightly and they separate.
2. Bright distractions at the edges, which pull the eye out of the picture. Crop them off or darken them.
3. Too much dead space with nothing happening in it, which usually means the subject is too small in the frame.

So what?: You can improve most pictures without redrawing anything, just by choosing a better rectangle.

Common misconceptions

• "The rule of thirds is a law of nature or a rule of mathematics." It comes from an English engraver's book of 1797 and is a useful default. Some people connect it to the golden ratio; the two are different numbers and the connection is loose at best.
• "A balanced composition is a symmetrical one." Symmetry is one kind of balance. Asymmetrical balance, where a large quiet area offsets a small intense one, is far more common in good pictures.
• "Composition is something you fix at the end." It is decided the moment you choose your frame and place the first big shape. Fixing it later usually means cropping, which is real, but the more expensive fix is redrawing.

The takeaway

Composition decides what the viewer sees first, what they see next, and whether they stay in the frame at all. The rule of thirds, from John Thomas Smith in 1797, says to place important elements on the third lines and the most important at a crossing point, and it works mainly because it prevents dead-centre placement and gives a picture direction. It is a default, not a law, and Leonardo and Raphael both built famous pictures on exact symmetry because symmetry delivers stillness and authority. Leading lines, framing, contrast, rhythm and asymmetrical balance are all separate tools you can reach for. And cropping with two L-shaped cards will improve more of your pictures than any other single habit in this course.

Sources

1. Wikipedia contributors. (n.d.). Rule of thirds. Wikipedia. en.wikipedia.org
2. Tate. (n.d.). Composition. Art Terms. tate.org.uk
3. Wikipedia contributors. (n.d.). The Last Supper (Leonardo). Wikipedia. en.wikipedia.org
4. Wikipedia contributors. (n.d.). The School of Athens. Wikipedia. en.wikipedia.org
5. Wikipedia contributors. (n.d.). Composition (visual arts). Wikipedia. en.wikipedia.org

Around 1490 Leonardo drew a man with four arms and four legs inside a circle and a square, and covered the page in mirror-written notes. The drawing is on paper, roughly thirty-four centimetres tall, and it lives in the Gallerie dell'Accademia in Venice, which shows it only rarely because light damages it.

Leonardo was testing claims made by the Roman architect Vitruvius about how a human body is proportioned: that the span of the outstretched arms equals the height, that the head fits a certain number of times into the body, and so on. He measured people and corrected some of it. The important part for you is the habit, not the conclusions. Proportion is a matter of comparison and measurement, which means it is learnable, which means the usual excuse does not apply.

The face everybody draws first

Draw a face from memory, quickly, before reading on. Then compare it to this list, which describes the drawing almost everyone produces:

• The eyes are near the top of the head, roughly a quarter of the way down.
• There is a large empty forehead above them.
• The nose and mouth are crowded into the lower third.
• There are no ears, or small ones stuck on near the top.
• The chin is small and the skull behind the face is missing.

This is not carelessness. It is a systematic error, and it happens because you are drawing what you pay attention to. When you look at a person, you look at their eyes and mouth, so those get enlarged in your mental image, and the forehead and skull, which you never examine, get compressed. Then your hand draws the mental image.

Which means the fix is not to try harder. The fix is to use measurements that overrule the mental image.

The one rule that fixes most of it

Here it is: the eyes sit halfway down the head. Not a quarter of the way down. Halfway between the top of the skull and the bottom of the chin.

Check it on a real person right now. Find a photograph of a face looking straight at the camera, hold a ruler or the edge of a piece of paper across the eyes, and compare the distance from there to the top of the hairline against the distance from there to the chin. They are close to equal, and if anything the top half is larger once you include the hair.

This one correction transforms most beginner drawings, because it gives the skull the space it actually occupies. Everything else on the head is a smaller adjustment.

Key idea: The eyes are halfway down the head. Almost every wrong face is wrong because they were put too high.

The rest of the map

Draw an oval slightly narrower at the bottom, like an egg standing on its narrow end. Then:

Feature	Where it goes
Eye line	Halfway between top of head and chin
Bottom of the nose	Roughly halfway between the eye line and the chin
Line of the mouth	Roughly one third of the way from the nose down to the chin, so nearer the nose than the chin
Ears	Top of the ear at the eye line, bottom at the base of the nose
Width of the head	About five eye widths across at the eye line
Gap between the eyes	Exactly one eye width
Corners of the mouth	Directly below the centre of each eye, or slightly narrower
Width of the nose	About the same as the gap between the eyes, at its base

The five-eye-widths rule is the one people find hardest to believe until they check it. Look at a straight-on photograph and count: eye, gap, eye, and then a gap the width of an eye on each outer side before you reach the edge of the head. Five. Nearly every beginner draws eyes too large and too far apart, which produces a face that looks vaguely amphibian.

These are averages, not commandments. Real faces vary in exactly these measurements, and that variation is what makes a specific person look like themselves rather than like a diagram. Use the map to get in the right neighbourhood, then look at your actual subject and note where they differ: this person's eyes are set unusually wide, that person's mouth is unusually close to their nose. Those deviations are the likeness.

Turning the head

Two things change when a head turns away from you, and only two.

First, the features crowd toward the far edge as the near side of the face takes up more space. Second, and this is the one beginners miss, the centre line of the face curves. Draw a light vertical line down the middle of your oval that bends around the form, the way a stripe painted on a ball bends. Every feature straddles that curved line: the nose sits on it, the mouth crosses it, and the two eyes sit either side at unequal widths.

The horizontal lines curve too. The eye line, nose line and mouth line all bow around the head like lines of latitude on a globe, and their curvature tells the viewer whether you are looking up at the person or down at them.

The body, measured in heads

The convenient unit for a figure is the height of its own head.

Figure	Approximate height in heads
Average adult	About 7.5
Heroic or idealised adult, used in comics and classical sculpture	8 to 8.5
Child around six years old	About 6
Baby	About 4
Fashion illustration	9 or more, with most of the extra length in the legs

The baby row explains a great deal. The reason a badly drawn child looks like a small adult is almost always that the head is too small. Young children have very large heads relative to their bodies, and shrinking the head by one unit ages a figure by years.

Two other useful landmarks: the elbows fall roughly at the waist, and the wrists roughly at the crotch. If your figure's arms end at mid-thigh, they are too long; if they end at the ribs, too short.

What matters here: Measure the figure in its own head heights, and check the elbow and wrist landmarks before adding any detail.

Breaking proportion, deliberately

None of this is a rule about how art should look. It is a description of how bodies are, and departing from it is a decision with consequences.

Caricature works by exaggerating the specific ways one face already differs from the average, so a long chin becomes longer and the likeness gets stronger rather than weaker. Ancient Egyptian relief and painting used hierarchical scale, where the pharaoh is drawn larger than his officials and they are drawn larger than labourers, so size encodes rank rather than distance. Fashion illustration stretches the figure to nine heads because clothes hang differently on a longer body.

In every case the artist knew the proportion and left it on purpose. That is the difference between a stylistic choice and a mistake.

The upshot: Learn the average so that your departures from it read as intention.

Common misconceptions

• "The eyes go near the top of the head." They go halfway down. This single error accounts for more failed portraits than everything else combined.
• "Beautiful faces follow the golden ratio." This claim circulates constantly and the evidence for it is thin. Face proportions do cluster around averages, and the golden ratio is a real number with real uses in mathematics, but the popular formula for a perfect face is not something you should draw from.
• "Proportion rules stop your work looking original." The rules put you in the right neighbourhood. Likeness and style both live in the deviations, and you cannot deviate accurately from something you cannot draw.

Summing up

Beginners draw faces wrong in a predictable direction, with the eyes too high and the skull too small, because attention shapes the mental image. The correction is measurement: eyes halfway down the head, nose halfway from eyes to chin, mouth a third of the way from nose to chin, ears spanning eye line to nose base, and five eye widths across the face. When the head turns, the centre line and the feature lines curve around the form like lines on a globe. Figures are measured in their own head heights, about seven and a half for an adult and four for a baby, and getting the head size wrong is why drawn children look like shrunken adults. Egyptian relief, caricature and fashion drawing all break these proportions on purpose, which is only possible if you know what you are breaking.

Sources

1. Wikipedia contributors. (n.d.). Vitruvian Man. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). Body proportions. Wikipedia. en.wikipedia.org
3. Tate. (n.d.). Portrait. Art Terms. tate.org.uk
4. Wikipedia contributors. (n.d.). Golden ratio. Wikipedia. en.wikipedia.org
5. Wikipedia contributors. (n.d.). Nefertiti Bust. Wikipedia. en.wikipedia.org

In 1916 the London Underground took delivery of a new alphabet. Frank Pick, the manager responsible for how the railway looked, had commissioned it from Edward Johnston, a calligrapher, and the brief was unusual: the letters had to be readable by a passenger moving past a sign at speed, in bad light, from a distance, possibly for only a second.

Johnston stripped off the decorative strokes on the ends of letters, built the shapes on simple circles and straight lines, and put a small diamond over the letter i. That alphabet is still in use on the Underground today, in an updated form, which makes it one of the longest-running pieces of graphic design anywhere.

Notice what the brief was not. Nobody asked for beautiful. They asked for readable in one second by someone who is not trying.

Serif, sans serif, and what the little feet were for

A serif is the small stroke finishing off the end of a letter's main strokes, the little feet on the bottom of an H in the type used in most novels. A sans serif face has none; sans is simply French for without.

The serifs are not decoration invented by a bored designer. They come down from Roman inscription carving, where a chisel finishing a stroke naturally produced a widening at the end, and they were kept because they help in long runs of small text by giving each letter a distinct outline and helping the eye track along a line.

Sans serif faces do better at a glance and at large sizes, which is why Johnston chose one for a railway. The rough working split is this: serif for long reading, sans serif for signs, headlines and screens. It is a tendency, not a law, and plenty of good design ignores it.

Remember: Serifs come from carving stone and help in long text. Sans serif is cleaner at a glance, which is why signs use it.

The parts of a letter you actually need

Term	What it is
Baseline	The invisible line all the letters sit on
x-height	The height of a lowercase x, and so of most lowercase letters
Cap height	The height of a capital letter above the baseline
Ascender	The part of a letter rising above x-height, as in b, d, h, k, l
Descender	The part dropping below the baseline, as in g, j, p, q, y
Counter	The enclosed white space inside a letter such as o, a or e
Stroke weight	The thickness of the lines making the letter

Two of these do more work than the rest. A large x-height makes a face look bigger and read better at small sizes, which is why signage faces tend to have generous ones. And the counters matter enormously: if you thicken your letters until the hole in the middle of an e closes up, the word turns into a row of blobs at exactly the distance you needed it to work.

Why round letters have to break the rules

Rule two guidelines and draw a capital H and a capital O between them, both touching top and bottom exactly. Step back. The O looks smaller.

It is not smaller; it is measurably the same height. But a flat top meets the guideline across its whole width while a curve only touches at one point, so the eye reads the round letter as falling short. Type designers fix this with overshoot: round letters like O, C, S and G are drawn slightly taller than flat ones, breaking through the guideline by a hair. Pointed letters like A and V get the same treatment.

This is the first real lesson of lettering, and it generalises. In type, what is measurably correct and what looks correct are often different, and looks correct wins every time.

Spacing, which is the whole job

Beginners think lettering is about drawing beautiful letters. Professionals will tell you it is about the gaps.

Kerning is adjusting the space between two particular letters. The reason it is needed is that letters have wildly different shapes. Put a capital A next to a capital V and their diagonals lean together, creating a large empty wedge. Put two capital H letters together and the two vertical strokes create a tight, dense gap. Space them at the same measured distance and the AV looks like a hole in the word.

So the rule is: space letters so the areas of white between them look equal, not so the distances are equal. That means moving A and V closer together and pushing H and H further apart, in defiance of the ruler.

Two checks that work without any training:

1. Squint at the word until the letters blur. Even spacing shows as an even grey stripe; a bad gap shows as a light hole.
2. Turn the paper upside down. You stop reading the word and start seeing the shapes, and uneven gaps become obvious.

The point: Equal gaps by measurement look wrong. Equal gaps by area look right, and squinting is how you check.

Hand-lettering a word, step by step

1. Rule four light guidelines: baseline, x-height, cap height, and one below the baseline for descenders. Use a ruler and press very lightly.
2. Sketch the word as skeletons: single-line letters with no thickness, just to get the shapes and rough spacing.
3. Check the spacing by squinting before you thicken anything. This is the moment to fix it, not later.
4. Thicken the strokes outward from the skeleton, keeping the weight consistent across every letter.
5. Let round and pointed letters overshoot the guidelines very slightly.
6. Add serifs, if you want them, last and identically on every letter.
7. Erase the guidelines and check the counters are still open.

What fails at a distance

• All capitals in long strings. Lowercase words have distinctive outlines, with ascenders and descenders sticking out, and readers use those shapes. Set in capitals every word becomes the same rectangle. Capitals are fine for two or three words on a sign and poor for a paragraph.
• Letters jammed together. Tight spacing is the single most common reason a hand-made poster becomes unreadable across a room.
• Low contrast. Yellow text on white, or dark grey on black, may look elegant on your desk and disappear at four metres.
• Decorative faces at small sizes. A gothic or brush script can work as a three-word headline and becomes noise in a paragraph.

Choosing a typeface on purpose

Type carries associations, learned from where you have seen it before. Helvetica, drawn in Switzerland in the 1950s and named after the Latin for Swiss, reads as neutral and corporate because it has been used on that kind of material for seventy years. A blackletter face reads as old or heavy metal, depending on the reader, for exactly the same reason: it is a memory of where you last saw it.

So the useful question when choosing type is not which one is nicest. It is what has my reader seen this shape on before, and is that the association I want.

Common misconceptions

• "A typeface and a font are the same thing." Loosely, in everyday speech, yes. Strictly, the typeface is the design, and the font is a particular version of it at a particular weight and size. It is the difference between a piece of music and one recording of it.
• "Serif type is old-fashioned and sans serif is modern." Both are old and both are current. Serif faces are still standard for long text in books and newspapers because they help the eye along a line.
• "Good lettering means drawing each letter beautifully." A word made of perfect letters with bad gaps looks worse than one with rough letters and even gaps. Spacing is what the reader actually sees.

Looking back

Johnston's 1916 alphabet for the London Underground was designed for one second of attention in poor light, and it is still doing that job. Serifs descend from Roman stone carving and help in long text; sans serif reads more cleanly at a glance and dominates signage. The parts of a letter worth knowing are baseline, x-height, cap height, ascender, descender and counter, and the counters are what close up and destroy readability when you overthicken. Round and pointed letters must overshoot the guidelines slightly or they look short, which is your first case of optical correction beating measurement. Spacing is judged by the area of white between letters rather than the distance, checked by squinting or by turning the page upside down. And a typeface carries whatever associations your reader has already built, which makes choosing one a question about them rather than about you.

Sources

1. Wikipedia contributors. (n.d.). Edward Johnston. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). Typeface. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (n.d.). Kerning. Wikipedia. en.wikipedia.org
4. Wikipedia contributors. (n.d.). Serif. Wikipedia. en.wikipedia.org
5. Wikipedia contributors. (n.d.). Helvetica. Wikipedia. en.wikipedia.org
6. Wikipedia contributors. (n.d.). Legibility. Wikipedia. en.wikipedia.org

In 1891 the Moulin Rouge in Paris hired Henri de Toulouse-Lautrec to advertise its star dancer. He produced a large colour lithograph, Moulin Rouge: La Goulue, and copies went up on walls around the city.

Look at how it is built. The dancer is a white shape in the middle, lit from below. In front of her, in the foreground, is a dark grey silhouette of another performer with no features at all. Behind her is a row of flat black cut-out shapes standing in for a crowd. There is no shading and almost no detail. The lettering is hand-drawn and irregular.

It works because a poster is not looked at. It is glanced at, by someone walking past, from four metres away, once. Everything in it is arranged for those conditions, and nothing is arranged for close inspection.

The three-second test

Every poster gets about three seconds. In that time the viewer will read, at most, three things, and they will read them in an order you either chose or left to chance.

So the first design decision is not a colour or a typeface. It is a ranking. Write down every piece of information the poster must carry, then number them 1, 2 and 3 by importance, and be ruthless: a school concert poster usually needs the event name first, then the date and time, then where to get tickets. The name of the deputy head who organised it is not in the top three.

Then make the design match the ranking. The number one item should be the biggest, boldest, most isolated thing on the sheet.

Why this matters: If everything on a poster is important, nothing is, and the viewer reads whatever happens to be largest.

Five ways to make one thing more important than another

Tool	How it works	Watch out for
Size	Bigger reads first; the strongest and bluntest tool	Three similar sizes read as one flat level, so make the jumps large
Weight	Bolder reads first at the same size	Bolding everything cancels the effect
Colour	One saturated colour in a quiet field pulls the eye	Two or three fighting accents cancel each other
Position	The top and the upper left get seen first in left-to-right reading cultures	Not universal; reading direction varies
Isolation	Space around a thing makes it important without changing its size	The most underused of the five

Isolation deserves the extra note. A modest line of text with a large empty area around it can outrank a much bigger line crowded by other elements. Space is a design tool, not leftover paper.

The grid, and what it is for

A grid is a set of invisible guide lines, usually margins plus a few columns, that everything on the page lines up with.

Its job is consistency. Without one, every element gets placed by a fresh judgement, and forty fresh judgements never agree with each other, so the page looks nervous. With one, the decisions have already been made and the eye reads calm.

A serviceable grid for a poster:

1. Rule a margin about one twelfth of the width in from every edge, and treat it as a wall nothing crosses.
2. Divide the remaining area into three equal columns with a small gap between them.
3. Place every element so its left edge, or its centre, aligns with a column edge.
4. Allow one element to break the grid deliberately, if you want it to shout. One.

Alignment: the thing that makes amateur work look amateur

Take any homemade poster and the fault will usually be the same. The title is centred, the picture is roughly centred, the date is centred but a bit off, and the ticket line is centred by eye and misses by four millimetres. Everything is nearly aligned to a centre line that does not quite exist.

Two fixes, either of which works:

• Align everything left. Pick a vertical line and start every block of text on it exactly. This creates a hard edge the eye can follow and looks deliberate immediately.
• Centre everything precisely, on a ruled line. Symmetry is fine, but it must be exact, because near-symmetry reads as a mistake while real symmetry reads as formality.

What fails is the mixture: two things centred, one thing left, one thing floating. Pick an edge and commit.

Worth holding on to: Every element should be aligned with something else on the page. If you cannot say what a block lines up with, it is in the wrong place.

Cutting information out: the Beck example

In 1933 Harry Beck, an engineering draughtsman, redrew the map of the London Underground. He threw away the geography: real distances, real curves, the true positions of stations. He drew the lines only at horizontal, vertical and forty-five degrees, spaced the stations evenly, and made central London larger than scale.

The result was geographically false and hugely more useful, because a passenger underground needs to know the order of stations and where to change, and does not need to know how far apart they are. Almost every metro system in the world now uses a diagram built on the same principle.

The design lesson is uncomfortable and worth taking: making something clearer usually means removing true information, not adding more.

Building the poster, step by step

1. List everything the poster must say. Rank the items 1, 2, 3 and then the rest.
2. Draw six thumbnails, each about the size of a matchbox, trying different arrangements. Thumbnails are fast because they are too small to fuss over.
3. Pick the strongest thumbnail and rule up a full-size sheet with margins and three columns.
4. Block in the number one element first, large, and give it space.
5. Add items 2 and 3, clearly smaller, aligned to the grid.
6. Add the picture or graphic, and check it does not compete with item 1.
7. Add the remaining details small, grouped together, in one place.
8. Take everything decorative out. Then take out one more thing.

Three tests before you call it finished

• The squint test. Half close your eyes. Whatever is still visible is your hierarchy. If the wrong thing survives, or if everything survives equally, fix the sizes.
• The six-pace test. Pin it up and walk six paces away. Item 1 should be readable. If it is not, the poster fails in the only condition that matters.
• The thumbnail test. Photograph it and shrink the photo to postage-stamp size. Anything that still reads is genuinely strong; the rest was decoration.

Bottom line: Test the poster in the conditions it will actually meet, which are distance, speed and indifference.

Common misconceptions

• "Empty space means I have not finished." Space is what makes the important element important. Filling every corner is the fastest way to make a poster unreadable.
• "More fonts make it more interesting." Two typefaces is plenty and one is often better. Interest comes from size and weight contrast within a family, not from variety of letterforms.
• "Centring everything is the safe option." Centring is only safe if it is exact and applied to everything. Approximate centring is the most common tell of an unplanned layout.

Recap

A poster is glanced at from a distance by someone who is not trying, which is why Toulouse-Lautrec built his 1891 Moulin Rouge poster from flat silhouettes and hand-drawn letters instead of detail. Design starts by ranking the information, because a design that treats everything as important leaves the viewer to pick. Size, weight, colour, position and isolation are the five tools for enforcing that ranking, and isolation is the one nobody uses. A grid of margins and columns replaces forty separate judgements with one, and every element should align with something. Harry Beck's 1933 Underground diagram shows the harder half of design, which is that clarity usually comes from removing true information. And three cheap tests, squinting, walking six paces back, and shrinking it to a thumbnail, will tell you whether any of it worked.

Sources

1. Wikipedia contributors. (n.d.). Henri de Toulouse-Lautrec. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). Poster. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. (n.d.). Harry Beck. Wikipedia. en.wikipedia.org
4. Wikipedia contributors. (n.d.). Tube map. Wikipedia. en.wikipedia.org
5. Wikipedia contributors. (n.d.). Grid (graphic design). Wikipedia. en.wikipedia.org

A Japanese woodblock print of the kind Hokusai designed was made by four people, and only one of them drew anything.

The artist produced the design on thin paper. A block cutter pasted that paper face down onto a block of cherry wood and carved away everything that was not a line, destroying the original drawing in the process. A printer inked the blocks, one for each colour, and rubbed the paper onto them by hand. A publisher paid for all of it, decided how many to make, and sold them.

That division of labour tells you something important about prints. A print is not a picture that happens to have been copied. It is a picture made through a machine, or a block, or a screen, and the properties of that intermediate object are what give the print its look.

The matrix, and what makes a print original

The block, plate or screen is called the matrix. You do not draw the picture. You make the matrix, and the matrix makes the picture, over and over.

This creates a distinction people get wrong constantly. An original print is one where the artist made the matrix as the artwork; every impression pulled from it is an original, not a copy, in the same way that every casting from a sculptor's mould is a sculpture. A reproduction is a photograph of an existing painting, printed. The first is art made in a printing process. The second is a picture of art.

Key idea: In printmaking you make the thing that makes the picture. Every good impression from the matrix is an original.

Four families, and how each holds ink

Family	Where the ink sits	Examples	Look
Relief	On the raised surface; carved areas print white	Woodcut, linocut, foam tray print	Bold shapes, strong blacks, visible cut marks
Intaglio	Down in lines cut or bitten into a metal plate	Etching, engraving, drypoint	Fine lines, dense detail, a plate mark pressed into the paper
Planographic	On a flat surface, held by grease and repelled by water	Lithography	Soft, drawn look, close to chalk or crayon
Stencil	Pushed through open areas of a mesh	Screenprint	Flat, even areas of solid colour

Relief is where you will start, because it needs nothing but a soft surface, something to draw with, ink and a spoon. It is also the family the Japanese woodblock tradition belongs to, and the one Hokusai's cutters were working in.

The mirror problem

Everything you cut prints backwards. The block presses onto the paper face to face, so left becomes right.

For a picture of a tree this rarely matters. For anything with text it matters enormously, and there is no way round it: letters must be cut in reverse. Write your word on paper, hold it up to a window or a mirror, and copy the mirrored version onto the block. Then check it again before you cut, because everyone gets this wrong once and only once.

It also affects composition. If your design has a figure looking to the left, the print will have them looking right. Decide which version you want before you start carving.

Making a foam tray block

A clean polystyrene food tray, the flat kind fruit or biscuits come on, is a genuinely good relief matrix. It is soft, it takes a line cleanly, and it costs nothing.

1. Cut the raised rim off the tray with scissors so you have a flat rectangle.
2. Draw your design on paper first, at the size of the tray. Keep it simple: bold shapes, thick lines, no fine hatching.
3. Lay the drawing on the foam and trace over it with a blunt pencil, pressing firmly. This transfers the design as grooves.
4. Remove the paper and deepen every line by drawing it again directly into the foam. Press hard enough to make a clear channel but not through the tray. A dead ballpoint pen works well.
5. Add texture with anything blunt: the end of a paperclip, a fork, a bottle cap pressed in.
6. Remember that every groove you press will stay white, and every uncut surface will print in colour.

So what?: In relief printing you are not drawing the image, you are drawing everything that will not be there.

Inking and pulling the print without a press

1. Squeeze a small amount of printing ink or thick acrylic paint onto a smooth surface such as a plate or an old tile.
2. Roll it out with a brayer, which is a small rubber roller, until the ink makes a light hissing sound and the roller is evenly coated. Too much ink fills your grooves; too little gives a patchy print.
3. Roll the ink onto the foam block in two directions. Do not press hard; you want a thin, even film on the surface only.
4. Lay a sheet of paper carefully on top, and do not move it once it touches.
5. Rub the back of the paper firmly and evenly with the back of a spoon, working from the centre outwards. Cover every part twice. This is called burnishing.
6. Peel one corner up to check. If it is patchy, lay it back down and burnish more; if it is good, peel the whole sheet off in one movement.

The second and third prints from a block are usually better than the first, because the first one absorbs the ink unevenly into a dry surface. Do not judge your block on impression one.

Reading a bad print

Fault	Cause	Fix
Grey and patchy overall	Not enough ink, or not enough burnishing	Re-ink and rub harder, longer, from the centre out
Lines filled in and lost	Too much ink, or grooves too shallow	Roll out more thinly; press the grooves deeper
Ghosting or doubled edges	The paper shifted during burnishing	Hold the paper still, or tape one edge down as a hinge
Dark smears in the white areas	Ink on the roller edge or on your fingers	Clean the edges of the block, and handle the paper by its corners
Blank patches at the corners	Uneven pressure while rubbing	Burnish right out to the edges, twice

A second colour, and why registration is hard

Printing two colours means printing twice, and the second impression has to land in exactly the right place relative to the first. Lining them up is called registration, and it is where multi-colour printing gets difficult.

The cheap method: tape a sheet of card to the table as a bed, draw the exact outline of your block on it, and draw the outline of your paper too. Put the block in its marked position every time and the paper in its marked position every time. Hinge the paper along one edge with a strip of tape so it lifts up and comes back down in the same place.

Print all of your first colour and let it dry completely. Then cut or make the second block, and print the second colour over the top. In the Japanese tradition, small carved marks called kento were cut into every block for the same purpose, so the printer could seat the paper identically each time.

How editions are numbered

A number written in pencil under a print, like 12/50, means this is the twelfth impression pulled from an edition of fifty. The letters AP mean artist's proof, an impression kept by the artist outside the numbered edition. When an edition is complete the matrix is often defaced or destroyed so no more can be made, which is what keeps the edition size honest.

The core of it: An edition is a promise about how many exist, and the numbering is how that promise is recorded.

Common misconceptions

• "A print is a copy, so it is worth less than a drawing." An original print is a work made in a printing medium, not a photograph of another work. Galleries collect Rembrandt etchings and Hokusai woodblocks as originals for exactly this reason.
• "You need a press." For relief printing you need a spoon. Hand burnishing is the traditional method in the Japanese tradition and works perfectly well on foam, lino and wood.
• "Whatever I draw is what will print." The image reverses, and in relief the areas you carve are the areas that stay blank. Both catch people out on their first block.

What to carry forward

A print is made through a matrix, and making that matrix well is the actual work; every good impression from it is an original rather than a copy. The four families differ in where the ink sits: on the surface in relief, down in the lines in intaglio, on a flat greasy surface in lithography, through a mesh in screenprinting. Relief is the accessible one, and a supermarket foam tray is a serviceable block. Everything prints in mirror image, which is survivable for a tree and fatal for a word. Ink thinly with a roller, burnish from the centre outwards with a spoon, and expect the second impression to beat the first. Faults are diagnosable rather than mysterious, and registration marks are what make a second colour possible.

Sources

1. Wikipedia contributors. (n.d.). Printmaking. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). Woodblock printing in Japan. Wikipedia. en.wikipedia.org
3. Tate. (n.d.). Woodcut. Art Terms. tate.org.uk
4. Tate. (n.d.). Linocut. Art Terms. tate.org.uk
5. Tate. (n.d.). Etching. Art Terms. tate.org.uk
6. Wikipedia contributors. (n.d.). Ukiyo-e. Wikipedia. en.wikipedia.org

In 1931 Marcel Duchamp visited Alexander Calder's studio in Paris and looked at some motorised constructions of wire and painted metal shapes. He suggested calling them mobiles, a French word meaning both moving and motive. The name stuck. Jean Arp then asked what Calder called the ones that did not move, and proposed stabiles.

Calder soon dropped the motors. The later mobiles hang from a ceiling and turn on air currents alone, which means the sculpture is never in the same configuration twice, and the artist is not in charge of what you see. That is an unusual thing to build on purpose.

This lesson is about the two things Calder was working with, which are also the two things you can work with on a kitchen table: what makes a three dimensional object read from every side, and how a flat material becomes a structure.

Sculpture is the art of the walk-around

A drawing has one viewpoint. A sculpture in the round has as many as there are places to stand, and it has to work from all of them, or at least it has to not fall apart from any of them.

The practical test is silhouette. Walk around a good sculpture and its outline against the background keeps changing into something readable and different: an arm appears, a gap opens, the mass shifts. Walk around a weak one and it collapses into a lump from three sides out of four. When you make something, turn it constantly. The view you are looking at while you work is the one view you can be certain about.

What matters here: Judge a sculpture by its silhouette from every angle, and turn the work as you build it.

Four ways to make one

Method	What you do	Materials	The catch
Carving, subtractive	Remove material until the form is left	Stone, wood, soap, plaster block	Nothing goes back on; a mistake is permanent
Modelling, additive	Build the form up from soft material	Clay, plasticine, papier mache, wax	Needs an armature or it sags
Casting	Make a mould, then fill it	Plaster, bronze, resin	Two stages, and the mould is the real work
Construction and assemblage	Join separate parts together	Wire, card, wood, found objects	The joins are the weak points and must be designed

Calder worked mainly in the last one: cut sheet metal, wire, and a great deal of thought about joints. Michelangelo worked in the first, and left several figures half emerged from their blocks, which is the clearest possible illustration of what subtractive means.

An armature is worth knowing about even at kitchen-table scale. It is an internal skeleton, usually wire, that a modelled figure is built around. Without one, a clay figure with thin ankles has to hold up a heavy body on wet clay, and it will not.

Relief: sculpture that stays on the wall

A relief is a sculpture that projects from a background rather than standing free. Low relief, sometimes called bas-relief, projects slightly, as on a coin. High relief projects a lot, with parts almost free of the background. Ancient Egyptian carvers also used sunk relief, where the figure is cut down into the surface rather than raised out of it, which reads sharply in strong sunlight.

Relief is the bridge between drawing and sculpture, and it is a good place to start, because you only have to make it work from one direction.

Why folded paper is stiff and flat paper is not

Hold a sheet of paper out horizontally by one edge. It flops. Now put a single fold along its length and hold it the same way, and it holds itself out almost straight.

Nothing was added. The paper is the same weight and the same material. What changed is that a flat sheet has almost no depth in the direction it needs to resist bending, and a folded sheet does. The fold creates a shape that has to be stretched or crushed before it can bend, and paper resists both.

This is the whole principle behind corrugated cardboard, folded steel roofing, and the ribs in a leaf. It is also why a paper structure that fails is almost always a structure with a flat panel in it.

Remember: Paper gets its strength from geometry, not from thickness. Fold it and it becomes a beam.

Three folds and what each one is for

• Accordion or fluting. Fold a sheet back and forth into narrow parallel pleats. This makes a panel that is very stiff across the pleats and still flexible along them. Stand it on edge and it will hold a surprising weight; a folded strip of copy paper can support a book.
• The tube. Roll a sheet into a cylinder and tape it. A tube resists bending in every direction at once, which a folded panel does not, and it is the strongest thing you can make from one sheet in ten seconds.
• Triangulation. A square made of four strips joined at the corners collapses into a parallelogram if you push it. A triangle cannot change shape without changing the length of a side. Bracing anything with triangles is why bridges and pylons look the way they do.

There are far more elaborate folds. The Miura fold, developed for packing large surfaces such as solar panels for spacecraft, opens and closes an entire sheet with one pull, because every crease is linked to every other. It is worth looking up simply to see how much behaviour can be designed into a piece of paper.

Two pop-up mechanisms

A pop-up is a paper machine: the energy of you opening a card is converted into a shape standing up.

The parallel fold, or step. On a folded card, make two parallel cuts perpendicular to the central fold, a few centimetres apart, cutting only through the fold and stopping partway. Push the strip between the cuts inwards while closing the card, so it folds the opposite way to the rest. Open the card and a step stands up. Anything glued to the front face of that step now rises when the card opens.

The V-fold. Cut a shape with a folded tab at its base, fold the shape down the middle, and glue the two tabs to the card so that the shape's centre fold sits across the card's centre fold at an angle. Opening the card pulls the shape upright into a V. This is what makes a pop-up dragon's head or a mountain rise toward the reader.

Two rules save most failed attempts. First, the mechanism must fold flat when the card closes, so every crease has to be planned. Second, glue only to the tabs; a single spot of glue on the wrong surface locks the whole thing solid.

Balance, and the mobile

Calder's mobiles look casual and are not. Every arm is a lever, and a lever balances when the weight on each side multiplied by its distance from the pivot is equal.

So a heavy shape close to the pivot balances a light shape far out. That single relationship gives you the design freedom that makes mobiles look the way they do: you are never obliged to hang matching objects, only to place them at matching moments.

Build one from the bottom up. Balance the lowest arm first, then treat that whole finished arm as a single weight hanging from the arm above it, and balance that, and so on upward. Trying to build downward from the top means every adjustment below ruins the balance above.

Common misconceptions

• "Sculpture means carving stone." Carving is one of four methods, and construction from cut and joined parts has been the dominant one for most of the last century. Calder's mobiles are cut sheet and wire.
• "A mobile balances when both sides weigh the same." It balances when weight multiplied by distance matches on both sides, which is why a small heavy shape can balance a large light one.
• "Thicker paper makes a stronger model." Up to a point, but geometry beats thickness every time. A tube of thin paper carries more than a flat sheet of card, and folding is free.

The short version

A sculpture in the round has to work from every angle, so silhouette is the test and turning the work is the habit. The four making methods are carving, modelling, casting and construction, and each has a characteristic failure: permanence, sagging, mould-making, and weak joints. Relief sits between drawing and sculpture and only has to work from one direction. Paper is stiff or floppy depending on its geometry, so fluting, tubes and triangles are the three folds worth knowing. Pop-ups convert the opening of a card into a standing shape through parallel folds and V-folds, and they only work if the whole mechanism can lie flat. And a mobile balances on weight multiplied by distance, built from the bottom upward, which is why Calder could hang a small dense shape against a large light one and have it hold.

Sources

1. Wikipedia contributors. (n.d.). Alexander Calder. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). Mobile (sculpture). Wikipedia. en.wikipedia.org
3. Tate. (n.d.). Alexander Calder 1898 to 1976. Artists. tate.org.uk
4. Tate. (n.d.). Relief. Art Terms. tate.org.uk
5. Wikipedia contributors. (n.d.). Miura fold. Wikipedia. en.wikipedia.org
6. Wikipedia contributors. (n.d.). Pop-up book. Wikipedia. en.wikipedia.org

In 1656 Diego Velazquez finished a large painting now called Las Meninas, which hangs in the Museo del Prado in Madrid. It shows a room in the Spanish royal palace. The five-year-old Infanta Margarita stands in the middle with her attendants. On the left, Velazquez has painted himself, standing at an enormous canvas whose front we cannot see. At the back of the room, a small mirror shows the king and queen.

Which means the painting refuses to answer the obvious question. Who is being painted on that hidden canvas? Are the king and queen standing where we are standing, and are we therefore in their place? Velazquez built a puzzle into a royal portrait, and it has kept people arguing for close to four hundred years.

This lesson puts pictures like that one next to pictures made under completely different rules, because the fastest way to understand any solution is to see another one.

Three problems, and nothing else

Strip away subject matter and almost every picture-maker in history has been working on some combination of three problems.

1. Depth. The surface is flat. The world is not. What do you do about that?
2. Likeness. How much of a specific person, or a specific thing, do you put in, and how much do you generalise?
3. Attention. A viewer will look at part of your picture. Which part, and how do you make that happen?

Different cultures answered these differently, and the differences are not stages on a road toward realism. They are choices made for reasons.

Four answers to depth

Tradition	The solution	What it gains, and gives up
Ancient Egyptian relief and painting	Figures arranged along horizontal bands, each shown from its clearest angle: head in profile, eye and shoulders frontal. Size shows rank, not distance	Gains total clarity about who is who and what is happening; gives up any sense of a single moment seen from one place
Japanese ukiyo-e prints, such as Hokusai's Great Wave	Flat areas of colour, strong outlines, overlapping shapes and steep diagonals; almost no cast shadow	Gains pattern, speed and enormous graphic power; gives up modelled solidity
Italian Renaissance painting, such as Masaccio's Holy Trinity in Santa Maria Novella	Linear perspective from a single fixed eyepoint, with everything measured back to a vanishing point	Gains a convincing continuous space; gives up the freedom to show more than one viewpoint
Cubism, from around 1907	Several viewpoints of the same object shown at once, in fragments	Gains information about the whole object; gives up the illusion of looking through a window

Look at that last row against the third. Cubism is not a failure to do perspective. It is a deliberate refusal, by painters who could do it perfectly well, on the grounds that a single frozen eyepoint is not actually how you know a jug on a table. You have walked around that jug.

The point: Flatness is not a stage before realism. It is a different answer to the same question, chosen for what it delivers.

Likeness, measured three ways

The painted limestone bust of Nefertiti in the Neues Museum in Berlin, made around 1345 BCE, is a portrait of an actual queen and is also perfectly smooth and symmetrical, generalised toward an ideal. It tells you what a queen is, more than what one woman looked like on a Tuesday.

Compare the Terracotta Army near Xi'an in China, buried with the first Qin emperor and rediscovered in 1974 by farmers digging a well. There are thousands of figures, and the faces vary: different hairstyles, different moustaches, different expressions, assembled from moulded parts with individually finished heads. The effect is of a real crowd rather than a repeated symbol, whether or not any single face portrays a specific soldier.

Then Rembrandt's Night Watch of 1642, in the Rijksmuseum in Amsterdam. It is a group portrait of a militia company, and Rembrandt refused to line them up in rows as such commissions usually did. He caught them mid-movement, half lit, some faces in shadow, one figure partly hidden behind a raised arm. Some of the men had paid to be in it and ended up in the dark.

Three approaches to the same problem: idealise, individuate, or dramatise. None is more advanced than another.

Colour, and two men who disagreed about it

Georges Seurat spent about two years on A Sunday on La Grande Jatte, finishing in 1886; it is in the Art Institute of Chicago and is roughly two by three metres. It is built from small separate touches of unmixed colour placed side by side, so that the mixing happens in your eye rather than on the palette. The technique became known as pointillism. Seurat was interested in colour theory as something close to a science, and the picture is calm, still and rigorously constructed.

Van Gogh painted The Starry Night in 1889; it is at the Museum of Modern Art in New York. It also uses separate strokes of strong colour, and it is doing the opposite thing. The strokes swirl, the paint is thick enough to cast its own small shadows, and the colour is pushed past anything you would see at night.

Same decade, same rough technique of broken colour, opposite purposes. One is a system for accurate optical mixing; the other is a way of getting feeling into paint.

Five questions that work on any picture

Describing art is a skill and it has a method. These five questions, asked in order, get you further than any amount of guessing about what the artist felt.

1. What is physically there? Name the objects, the people, the colours, the size, the material. No interpretation yet. Just inventory.
2. Where does my eye go first, and second? Then find the cause. Contrast, size, an isolated shape, a line pointing at it, faces looking at it.
3. How is it arranged? Symmetrical or not, what is central, what is cropped, where is the horizon, what is repeated.
4. How does it handle light and depth? One light source or many, gradual shading or flat colour, perspective or overlap or nothing.
5. What is it doing to me, and which of the above is causing that? This is the only step where you talk about feeling, and it has to be tied to something you named in steps 1 to 4.

Worth holding on to: Say what is there before you say what it means, and connect every claim about meaning to something visible.

Running the method on Las Meninas

Question 1: a large room, nine people and a dog, a huge canvas seen from behind, a mirror on the back wall, an open doorway with a man on the steps, dark browns and greys with the child in pale silver.

Question 2: the eye goes to the Infanta, because she is the lightest thing in a dark room and the two attendants are turned toward her. It goes second to the bright doorway at the back, which is the only other strong light.

Question 3: the child is roughly central and low; the composition is not symmetrical; the enormous canvas cuts off the left edge; the ceiling is high and largely empty.

Question 4: light comes from a window on the right and from the doorway behind, the space recedes convincingly, and the far wall is a real wall you could pace out.

Question 5: it feels like an interrupted moment rather than a posed portrait, and that comes from the light on the child, the turned heads, the mid-step figure in the doorway and the fact that half the people are looking out at us rather than at each other.

Notice that nothing in that description required knowing who anyone was.

Common misconceptions

• "Art got better over time as artists learned realism." Egyptian artists were not attempting perspective and failing. They had a system that encoded rank and identity, and they used it consistently for three thousand years. Later painters gave up things earlier ones had.
• "Modern art is what happens when you cannot draw." Picasso's academic drawings from his teens are extremely accomplished. The break with perspective was a decision made by people who could do the alternative.
• "You need to know the story behind a painting to talk about it." Background helps, but the five questions run entirely on what is visible, and describing what is there is a stronger starting point than repeating a fact you read on a label.

Where this leaves us

Every picture-maker is solving depth, likeness and attention, and the differences between traditions are choices rather than a ranking. Egyptian relief bought clarity by giving up a single viewpoint; ukiyo-e bought graphic power by giving up modelling; Italian perspective bought continuous space by fixing the eye in one place; Cubism gave that up again to show more than one view at once. Likeness can be idealised as in the Nefertiti bust, individuated as in the Terracotta Army, or dramatised as in Rembrandt's Night Watch. Seurat and Van Gogh used broken colour in the same decade for opposite reasons. And five questions, asked in order and starting with plain inventory, will get you further into any picture than guessing at what the painter was feeling.

Sources

1. Wikipedia contributors. (n.d.). Las Meninas. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). A Sunday Afternoon on the Island of La Grande Jatte. Wikipedia. en.wikipedia.org
3. Rijksmuseum. (n.d.). The Night Watch, Rembrandt van Rijn, 1642. Rijksmuseum Collection. rijksmuseum.nl
4. Wikipedia contributors. (n.d.). Terracotta Army. Wikipedia. en.wikipedia.org
5. Tate. (n.d.). Cubism. Art Terms. tate.org.uk
6. Wikipedia contributors. (n.d.). Pointillism. Wikipedia. en.wikipedia.org

When J. M. W. Turner died in 1851 he left the contents of his studio to the British nation. That bequest included something like three hundred sketchbooks, along with tens of thousands of drawings and watercolours, and most of it is now held by Tate.

The striking thing about the sketchbooks is how unfinished they are. Pages of coastline reduced to three pencil lines. Colour notes that look like smudges. The same castle drawn six times because he had not got it yet. These were not made to be looked at. They were the machinery behind the finished paintings, and the machinery was enormous.

This last lesson is about that machinery, and about the other thing that makes people improve, which is being able to hear and give a useful judgement.

What a sketchbook is for, and what it is not

A sketchbook is not a portfolio. A portfolio is your best work, chosen and presented for someone else. A sketchbook is a workshop, and it is supposed to be full of things that did not come off.

This distinction sounds obvious and is broken constantly. Almost every beginner turns their sketchbook into a display case, drawing carefully, only attempting subjects they already handle well, and stopping the moment a page goes wrong. The result is a tidy book and no progress, because you cannot improve at things you refuse to attempt badly.

Key idea: A sketchbook is where you are allowed to be bad. If every page is presentable, it is doing the wrong job.

Four kinds of page worth having

Page type	What goes on it	What it trains
Observation	Whatever is actually in front of you, drawn from life	Looking, which is the skill everything else rests on
Study	One difficult part, repeated: eight hands, six eyes, the same shoe from four angles	A specific weakness, deliberately
Notes and thumbnails	Tiny compositional sketches, written ideas, colour notes	Planning, and remembering ideas you would otherwise lose
Copies	Drawings made from other artists' work, labelled as such	Technique, by reverse engineering someone else's solution

The copies row needs a word of caution. Copying to learn is an ancient and completely respectable practice, and copying an artist's work and presenting it as your own is not. Write on the page whose work it is. That one habit keeps the distinction clean, and it becomes citing your sources when the work goes public.

How to keep one so that you use it

• Small enough to carry. A book that lives in a drawer is a book you never open. Postcard sized is enough for most people.
• Set a floor, not a target. One drawing a day, however small, beats an hour every Sunday that you skip twice a month. Five minutes counts.
• Date every page. This costs three seconds and turns the book into evidence. In six months you will want to compare.
• Never tear a page out. This is the important one, and it deserves its own paragraph.

Tearing out the bad pages destroys the record of what you can now do and could not do before, which is the only reliable proof of progress you will ever have, and it teaches you that failed drawings are shameful. Both consequences make you worse. Draw a line through it if you must and turn the page.

Critique: not an opinion, a procedure

Most feedback on art is useless, because it arrives as a verdict. I like it. It's good. Something looks off. None of these tell the maker what to change.

Useful critique follows an order, and the order is what makes it useful. Say what is there, say how it works, say what it seems to be about, and only then judge.

1. Describe. What is on the page. Objects, colours, medium, size. No judgement at all, and this step is not optional; it is where you notice things.
2. Analyse. How it is put together. Where the eye goes, how it is composed, how the values sit, what the marks are doing.
3. Interpret. What it seems to be about or aiming for, tied to something you named in the first two steps.
4. Judge. How well it achieves that, and what specifically would move it closer.

Notice that judging comes last and is measured against the work's own aim rather than against your taste. A deliberately flat, graphic poster is not failing at shading. It is not attempting shading.

In short: Describe, analyse, interpret, then judge. A verdict without the first three steps is not usable by the person who made the thing.

Three sentence-starters that turn comments into changes

• What is working is ... because ... The because is compulsory. Nice colours is worthless; the blue background makes the orange figure jump forward is something the maker can repeat on purpose.
• My eye goes first to ... and I think that is because ... This tells the maker whether their hierarchy is doing what they intended.
• One thing I would try is ... Framed as an experiment rather than a correction. It leaves the decision with the person who made it, which is where it belongs.

Notice what is missing: any sentence beginning with I like or I don't like. Preference is real but it is not information.

Seeing your own work, which is the hard part

After an hour on a drawing you can no longer see it. Your brain has merged what is on the paper with what you intended, and the intention is filling in the gaps. Every one of these tricks works by breaking that merger.

• The mirror. Hold the drawing up to a mirror. The reversal makes it a new image, and errors in symmetry and proportion, particularly in faces, appear instantly.
• Upside down. Turn it over. You stop reading it as a face or a house and start seeing shapes and values, which is what a stranger sees.
• Distance and thumbnail. Pin it up and walk away, or photograph it and shrink the photo. Composition and value problems survive shrinking; detail does not, which is exactly why this test is honest.
• Time. Put it in a drawer for a day. This is the most effective one and the least popular.
• Squint. Half close your eyes and see whether the big shapes still read.

What to do with what you find

Change one thing at a time. If a critique produces five points, fix the biggest and look again, because fixing the biggest often dissolves two of the others.

And when a drawing is badly wrong in its proportions or its composition, redraw rather than repair. Beginners spend an hour erasing and re-erasing a head that is the wrong shape, when drawing it again from the start takes fifteen minutes and comes out better, because the second attempt begins with everything the first one taught you.

Bottom line: Fix the largest problem first, and start again rather than repairing something that is structurally wrong.

Common misconceptions

• "A sketchbook should look good." A sketchbook full of successes is a sketchbook where nothing hard was attempted. Turner's are full of three-line coastlines and repeated failures.
• "Critique means finding faults." The first three steps of a critique contain no judgement at all, and naming what works, with a reason, is what lets someone do it again on purpose.
• "If I just look harder at my drawing I will see what is wrong." You will not, because familiarity has merged the drawing with your intention. You need a mirror, a shrunken photo, or a day away from it.

Looking back

Turner left roughly three hundred sketchbooks of unfinished, repeated, provisional work, and that is what a sketchbook is: a workshop, not a display case, kept small enough to carry, dated, and never torn. Fill it with observation from life, studies of the specific thing you are bad at, thumbnails and notes, and labelled copies from other artists. Critique is a procedure in four steps, describe, analyse, interpret, judge, and only the last of those is an opinion, measured against what the work was trying to do. Feedback becomes usable when every observation carries a because. And since you cannot see your own work after an hour with it, use the mirror, the upside-down view, the shrunken photograph and a night's delay, then change one thing, starting with the largest.

Sources

1. Tate. (n.d.). Joseph Mallord William Turner 1775 to 1851. Artists. tate.org.uk
2. Wikipedia contributors. (n.d.). Sketchbook. Wikipedia. en.wikipedia.org
3. The J. Paul Getty Museum. (n.d.). Elements of art and formal analysis. Getty Education. getty.edu
4. Tate. (n.d.). Art terms glossary. Tate. tate.org.uk
5. Feldman, E. B. (1994). Practical art criticism. Prentice Hall.