1. The idea
Books on menswear mostly teach by example: this man in this jacket with this tie. That helps only as far as you resemble the man in the picture. Underneath the examples sit a few rules that are well understood and can be measured. Valet measures you, applies the rules and tells you why each verdict holds, so you can apply the reasoning yourself.
Every rule carries an evidence badge:
- A Perception science. Measured and replicated: colour difference, simultaneous contrast, the visibility of fine detail.
- B Convention. Long-standing, widely agreed practice in tailoring and colour theory. Stable across sources, but not experiments.
- C Popular system. Widely used but not validated, such as "seasonal" colour analysis. Valet does not use these to score you.
Most rules have two halves: a quantity that can be measured (A) and a judgement about what looks good at a given value (B). Where a rule shows both badges, the measurement is science and the threshold is convention. Every number that is a threshold or a weight is a starting value, written down so it can be argued with.
Rules can be broken on purpose. A break reads as intentional when it is repeated and anchored by a calm base; a later module (Outfit check) will model this.
2. Measuring colour from a photo
A camera records colour under whatever light falls on you. Warm indoor light makes everyone look golden; a cloudy window makes everyone look grey. So Valet asks you to sample a sheet of white paper in the same photo.
A The correction is a von Kries adaptation: the sample is converted to cone-like signals (the Bradford transform used in colour management), each signal is scaled so that the paper becomes a neutral white at CIELAB L* ≈ 95, and the result is converted back. The same scaling is applied to your skin, hair and eye samples, which also normalises exposure. Each sample is averaged over a small patch, in linear light, to suppress noise.
Colours are then converted from sRGB to CIELAB (CIE 1976, D65 white): L* is lightness (0 black to 100 white), a* runs green to red, b* blue to yellow. Chroma C* is the distance from grey, and the hue angle h is the direction.
A colour you set by hand, with the tone ladder and sliders, is taken as the true colour and is not corrected.
- Phone cameras process images: HDR, beauty filters, "warmth", and a tone curve that makes "linear light" only approximately linear. Daylight, no filter and paper close to the face give the best result.
- Office paper contains optical brighteners that glow slightly blue in daylight. Treating that paper as neutral nudges every reading a little warm. This is one reason the temperature reading is never given more than medium confidence unless it is far from neutral.
- Paper that is blown out to pure white, or in shadow while the face is lit, cannot be used; Valet warns when the sample looks like either.
3. The four dimensions of your colouring
- Contrast A B
- The strongest lightness step (ΔL*) between your skin and a feature: hair, eyes, and the whites of the eyes. Below 25 is low, 25–45 medium, above 45 high. Lightness contrast between features and skin is a studied property of faces (Russell 2009); using it to choose clothes is the tailoring convention (Flusser 2002). Three refinements, all working values:
- Hair. Hair frames the whole face, so its step counts in full. With no hair, the same colour survives only in the brows, a thin line; that step counts at 80%.
- Eyes. The iris is the smallest feature, so its step counts at 60%. The writers rank hair first, then skin, then eyes (Permanent Style 2019), and treat the eyes as nudging up or down a contrast that hair and skin have already set. Counted in full, a fair man with white hair and dark eyes read as high contrast, which every source contradicts.
- Whites of the eyes. Not sampled; assumed at L* 80 and counted at 80%. On fair skin this adds nothing. On deep skin it is the strongest light-dark step in the face, and leaving it out would wrongly read deep skin with dark hair as low contrast.
- Depth A
- The average lightness of what is seen, weighted by area: skin 1, eyes 0.08, and hair 0.08 (brows only), 0.45 (short) or 0.85 (long). A fair, bald head is light; the same face under long dark hair is medium.
- Temperature lean B
- The hue angle of your skin in CIELAB. Skin hues sit roughly between 45° (pinker, cooler) and 70° (more golden, warmer); Valet maps 57° to neutral and ±8° to fully cool or warm. This is the least reliable of the four. The camera limits above act on exactly this number, and skin hue also moves with tanning and with depth, so a single neutral point for everyone is crude. It carries a confidence label, and verdicts that rest on a low-confidence reading say so.
- Chroma A
- How vivid your own colouring is: the plain mean C* of your features. Unweighted on purpose: the eyes are small, but they are where people look.
Names instead of numbers
B You can set each feature by name. A name stands for a typical colour, so the reading is coarser than a good photo but far better than a guess on three sliders, and it uses words you already know. The sliders remain under "Fine-tune the colour" for anyone who wants them.
- Skin has two parts, as the guides say (Gentleman's Gazette 2018): how light or dark it is, and its undertone. Valet offers ten depths from very fair to very deep and three undertones. Pink sets the skin's hue 7° below Valet's neutral point and golden 7° above, which reads as cool and warm; "neutral or not sure" sits on it. The home tests for undertone (wrist veins, white paper in daylight, whether you burn or tan) are in the "?" beside the control. Their own author calls the vein test "not hard science", and the burn-or-tan question really measures something else: it is the basis of the dermatologists' Fitzpatrick phototypes (1975), which describe reaction to sun, not colour. Skin colour itself is classed in dermatology by the individual typology angle, computed from the same L* and b* Valet measures (Chardon and others 1991), and in imaging by the ten-tone Monk scale (2023).
- Hair uses the hairdresser's chart, the one naming system for hair that people meet in daily life: ten depth levels from 1 (black) through the browns (2 to 5) to the blonds (6 dark blond, 7 blond, up to 10, the lightest), a tone (ash, natural, golden), and the natural reds (auburn, copper red, strawberry blond). Anthropology has its own charts, the Fischer–Saller scale for hair and the Martin–Schultz scale for eyes; the everyday words cover the same ground.
- Eyes run from blue through grey, green, hazel and amber to dark brown.
When a colour comes from a photo or from the sliders, Valet shows the closest name beside it, so you can see where the measurement lands in ordinary words.
Winter and summer
A B Skin is the one part of your colouring that changes through the year. Sun makes it darker and a little redder. The best measurement Valet could find followed 103 students through two weeks of outdoor sun in late summer: on the forehead, cheek and neck, lightness fell by 2.5 to 2.9 L*, red rose by up to about 1 a*, yellow hardly moved, and the hue angle shifted by under 2.5° toward red (Jiang et al. 2022). That is the A part: the direction and size of a light tan.
So Valet keeps two skins. Winter is your base profile. Summer is one of three:
- An estimate, the default: your winter skin with the measured two-week change applied. It is a light tan. If you tan more, it underestimates.
- Same as winter, for people who hardly tan.
- Your own: the depth you reach by late summer, picked by name, or skin sampled from a summer photo with its own sheet of white paper. Hair and eyes keep the correction from your first photo.
Keep your winter undertone when you pick a summer depth. The hue barely moves with a tan, so pink skin stays pink, only darker. Choosing "golden" for summer because the skin looks browner would wrongly turn your palette warm.
Every rule then runs on the summer skin, and the effects follow from the arithmetic rather than from separate summer rules (B):
- Pale colours separate better. White, ivory and écru sat close to fair winter skin; against a tan they stand apart, so they often move up a tier. This matches the writers' remark that richer, deeper colouring carries pastels and bright white (Barker 2020).
- Contrast to dark hair falls; to light hair it rises. Darker skin closes the gap to dark hair and opens it to blond. (Sun also lightens hair, which Valet does not model.)
- Colours near the tan itself lose ground. Camel, sand and some pinks can come too close to tanned skin.
There is one experiment on exactly this question. Perrett and Sprengelmeyer (2021) found observers chose cooler blues to go with fair skin and warmer orange-reds with tanned skin. Valet's tan estimate does not make skin warmer by hue, so that preference reaches the scores only through depth and separation. The temperature rule stays at B.
The profile card shows both seasons side by side and lists every colour whose tier moves. The switch at the top of the page decides which season every tab shows.
What the seasons change besides skin, and Valet does not model. The light and the surroundings change too. In winter, at least in northern Europe, the sun is low and weak, the sky is often overcast and streets, buildings and bare trees are grey; in summer the sun is high and strong and the surroundings are green and bright.
- A Brighter light makes colours look more colourful and contrasts look stronger (the Hunt and Stevens effects, both part of standard colour-appearance models such as CIECAM02). Under a grey winter sky the same cloth looks quieter and darker; in summer sun it looks louder, and strong contrasts look harder.
- B Surroundings set what stands out. Against grey winter streets a coloured piece is the brightest thing around; against summer green and strong light, pale and light colours look fresh and very dark ones heavy.
- Why the tiers do not move for this. Your face and your clothes are lit by the same light and seen against the same surroundings, so how a colour sits against your skin, which is what the tiers measure, changes far less than how the colour looks on its own. What changes is the overall volume: a vivid colour that is right in winter can be loud in July, and a pale outfit that is fresh in July can look washed out in February. That is the reason for the daylight test: judge cloth in the light, and the season, you will wear it in.
Grey hair
A B Greying hair is pigmented strands and white strands side by side. A hair is under a tenth of a millimetre wide, far finer than the eye resolves at talking distance, so the head reads as one blended colour: the same fact that turns a fine stripe into a solid (section 11). That much is perception science. How light the blend is, is less certain. Flat patches would average in linear light, but hair is a translucent mass in which dark strands shadow white ones, and plain averaging makes a quarter-grey head far lighter than it looks. Valet blends your pigmented colour with white hair in CIELAB, which falls between the two. It is a working model.
You give the colour your hair has where it still has pigment, and the share of grey. The blend then feeds every reading, and three things follow:
- Contrast moves, and the direction depends on your skin. On lighter skin grey closes the gap between hair and skin, so contrast falls: dark brown hair on fair skin reads high, at half grey medium, and black beside a white shirt stops being a safe choice. On deep skin grey opens the gap, so contrast rises. This is what the writers say in words: a fair man with greying hair is low contrast, a dark-skinned man with greying hair high contrast (Gentleman's Gazette 2019; The Rake 2020).
- Blond hardly changes. Blond and grey are about equally light, so the step to the skin barely moves. The same guides put blond and greying hair in one group for this reason.
- Colouring becomes more muted. Grey has almost no chroma, so your own chroma falls and very vivid colours overpower you sooner.
Limits. Grey rarely arrives evenly; temples and beard usually go first, and Valet uses one figure for the whole head. Up close, salt-and-pepper hair is itself a fine pattern, which the score ignores. Separately from hair, the contrast between features and skin falls with age and is one of the cues by which age is judged (Porcheron, Mauger and Russell 2013, measured on women's faces), so a reading taken ten years ago is worth repeating.
4. Contrast matching
A B The eye reads a face as balanced when the contrast of what surrounds it is in the same range as the face's own contrast. If the clothes near the face are much more contrasty than the face, they take the attention; if they are much less, the face loses definition. Valet applies this in two places.
A colour against your skin. Each colour is compared with a band around your face contrast:
lower = max(8, 0.5 × contrast) upper = 1.5 × contrast + 15
The lower bound is about definition, which can come from lightness or from hue, so it is tested against the larger of ΔL* and 1.2 × ΔE2000. The upper bound is about heaviness, which is lightness alone, so it is tested against ΔL*. Inside the band scores fully; outside, the score falls off over 20 units. The profile card draws this band in green on your lightness scale.
Shirt against jacket. The lightness chart and the Outfit check check that the two layers at the face differ by at least the lower bound, so the face has a frame. It only objects to a very strong step when your own contrast is low.
Limit. The convention was written from mostly light-skinned examples. The whites-of-the-eyes term is Valet's own way of correcting that. The literature reaches the same place by assertion: men with dark skin are placed in the high-contrast group, with a navy suit and white shirt as the example (Gentleman's Gazette 2018; Permanent Style 2019).
5. Separation from the skin
A B When a colour is close to your skin in lightness and hue, face and garment separate weakly. Valet measures the distance with CIEDE2000, the standard formula for perceived colour difference, and scores it from 0 at ΔE 10 to full at ΔE 20.
Those two numbers are convention, not science. A ΔE of 10 between two flat patches is plainly visible; the claim is only that, across a soft boundary with shadow and texture and at talking distance, differences in this range give the face a weak edge. CIEDE2000 was also built for small differences and is less exact for large ones. Beige, camel and oatmeal usually fail here for fair skin.
So does white, a little: on fair skin it sits about as far from the face as écru does, and lands in tier 2. Its usual job is to be the light half of a light-dark step under a dark jacket, which section 4 scores separately.
6. Simultaneous contrast (Chevreul)
A B In 1839 the chemist Michel Eugène Chevreul, working on dyes at the Gobelins tapestry works, showed that two colours placed side by side each push the other toward its opposite. A green collar makes the skin above it look redder; a yellow one makes fair skin look bluer, which reads as grey or tired; blue and violet push skin slightly warmer.
The direction of the push is science. Its size on a real face is not well measured, and whether a given push flatters is a judgement. Valet estimates it by moving your skin's a* and b* one tenth of the garment's own a* and b* in the opposite direction, then checks whether the result emphasises something unwanted: ruddiness, a grey cast or sallowness.
A second, physical effect works the other way: light reflected off the garment tints the jaw and neck with the garment's own colour. Valet does not model it.
A B Redness (set on the You tab). A red or pink beside a ruddy skin is echoed by the red in it and picks it out; green, the complement of red, makes the skin look redder by simultaneous contrast. Colour analysts give the same rule as "avoid the complement of the colour you want to play down" (NMSU Extension, Guide C-315). Both effects grow with saturation and shrink for dark, muted colours; blues and greys are the calm choice.
7. Temperature and chroma
B Colours are warm (reds, oranges, yellows, olive, camel) or cool (blues, blue-greens, violets), with blue-reds such as burgundy and rose near neutral. Valet derives this from the hue angle and weakens it for greyish colours. Near the face, a colour's temperature should agree with your lean; if your lean is neutral, only strong extremes are penalised.
There is one direct experiment. Perrett and Sprengelmeyer (2021) had observers choose clothing colours for photographed faces and found cooler blues preferred for fair skin and warmer orange-reds for tanned skin. That supports matching temperature to skin, but it compared fair with tanned skin in white women, not pink with golden undertone, so the rule stays at B.
B A colour much more saturated than your own colouring (more than about 35 C* above it) arrives before you do. The penalty grows with the excess.
B Metal frames are a second temperature at the face. Gold frames with a warm colour count against the colour unless your own lean is warm, where the two agree; silver frames with a cool colour count against it only if your lean is warm.
8. From scores to tiers
The rules combine into one score at the face. The weights are a judgement about relative importance, not a measurement:
0.30 contrast + 0.20 separation + 0.25 temperature + 0.25 chroma − Chevreul and frame penalties
A colour is also only as good as its worst failing, so each tier has a floor for the weakest rule:
- Tier 1 — anywhere: score ≥ 0.80 and no rule below 0.70. The base of a wardrobe.
- Tier 2 — deliberate colour: score ≥ 0.62 and no rule below 0.45. Fine at the face as the one chosen colour of an outfit.
- Tier 3 — below the belt: fails the above but is quiet (C* under 30) or fails mainly by blending with the skin. Trousers, shoes and belts.
- Tier 4 — never at the face: fails and is vivid. Small accents, or one bold piece below the belt.
The colour map on the Palette tab runs the same score over every hue and lightness at one strength. Hue and lightness there are OKLCH, so neighbouring squares are even steps to the eye, and squares your screen cannot show are left empty. The size of a square is its verdict, so the shape can be read without judging colour: large for tier 1, large with a white dot for tier 2, small for tiers 3 and 4. The Strength slider sets the saturation of the whole map; moving it shows how far toward vivid your palette reaches.
9. Pairing
B Pairings use OKLCH (Ottosson, 2020), a colour space in which equal steps look roughly equal, so hue relationships are more trustworthy than on the old artist's wheel.
- Neutrals (OKLCH chroma below 0.04) pair with anything, provided the lightness differs clearly. When the base itself is a neutral, any hue is free.
- Analogous colours lie within about 35° of hue: calm and safe.
- Complementary colours lie about 180° apart (±30°): lively, best when one of them is dark or muted. These are perceptual opposites, which differ from the painter's wheel: the opposite of blue is a yellow-orange, and the opposite of red is a blue-green rather than green.
- Any pair needs a lightness difference of at least about 12 L*, or the two read as one blur.
Navy, camel and tan are neutrals in the wardrobe sense but not by this measure; they carry enough colour to have a hue, and are paired as colours.
A useful outfit template: two neutrals and one colour.
Build around one colour
B Tailors and shops once chose a colour for the season and built everything around it. On the Palette tab you can do the same for yourself: choose one colour, and Valet says where it can go on you (its tier), and why when it is weak. For a colour that does not suit you near the face it offers its nearest versions that do (mustard leads to sage and moss), and it refuses the face outright for a tier 4 colour, allowing at most a small accent far from it. Around the colour it lists neutrals that suit you and separate clearly from it, colours that relate to it as neighbours or opposites and suit you, and colours to keep apart from it: near matches and colours that compete. "Put it in the study" puts the colour where it belongs, with a neutral and, where it helps, an accent chosen to go with it.
10. The outfit as a whole
B Once there is a whole outfit, rules apply between pieces. These are the Outfit Check card's questions, made measurable. All are convention unless badged otherwise.
- Positions, not fixed garments. An outfit is a set of positions: a hat, next to the neck, neckwear, a mid layer, an outer layer, a scarf, a pocket square, trousers, socks, shoes and belt. Only next to the neck, trousers and shoes are needed, so a summer shirt and trousers is a whole outfit. Each position holds a garment type: a shirt, roll neck, polo or T-shirt; a tie, bow tie or cravat (ascot); a crew neck, V-neck or cardigan; a jacket, overshirt or coat. Neckwear sits at the collar and needs one; the scarf is its own position, worn over everything, so a tie and a scarf can go together under a coat. The type carries what the rules need: whether it frames the face, which patterns are used on it, and the largest pattern scale that suits it (section 11).
- At the face. Every layer above the waist, and the pocket square, is scored at the face, using the colour it adds up to from a distance (section 11 for patterns). Each should be tier 1 or 2.
- Something frames the face. The Outfit Check's third question: name the thing that frames the face. A shirt or polo collar, a roll neck, a shawl collar, lapels, a coat collar or a scarf all count. A T-shirt or crew neck alone does not, which is noted as casual rather than failed. A tie needs a shirt collar to sit under; a V-neck knit is cut to show one.
- Light-dark step A B. Somewhere among the layers at the face there should be a step at least the lower bound of section 4. With a single layer, the collar alone frames the face, which is noted but not failed.
- One temperature above the waist. A warm piece among cool ones, or the reverse, reads as the odd one out.
- Colours that compete. Wardrobe neutrals (OKLCH chroma under 0.06) are free. Coloured pieces are grouped into families within 35° of hue. One family is the classic template. Two families are fine when they are complementary (best if one is dark or muted) or when one is only a small accent (tie, shoes, a pattern's second colour). Two families that are neither neighbours nor opposites compete. Three or more is flagged.
- Near match A B. Two pieces between ΔE2000 1.5 and 7 apart, such as navy with midnight, are close enough to look intended to match and far enough to fail. Below 1.5 they match: a jacket and trousers that match read as a suit.
- Top and bottom differ by at least 12 L*, or the halves blur.
- One statement. A tier 2 colour is "the one chosen colour of an outfit" (section 8). A second, different tier 2 colour at the face competes with it, and so does a tier 4 colour below the belt, which is the bold move of a bold day. Two statements are flagged as a caution, three as a failure. Neutrals do not count, and two pieces in one colour count once.
- Tie against shirt. The tie should separate from the shirt; darker than the shirt is the reliable order.
- One colour, one piece. A colour works as one piece among neutrals (rule 4). The same colour, or one colour family, over the top and the trousers reads as a uniform. Wardrobe neutrals here are the menswear ones (navy, greys, black, white, creams and browns), so tonal charcoal or all-black evening dress passes, but a forest shirt over forest trousers does not, even though forest is as muted as navy. A coloured suit, jacket and trousers cut as one, is the exception.
- The card's order. The checks are grouped as the Outfit Check card asks them: 1 temperature, 2 contrast, 3 frame, 4 sheen, 5 pattern, 6 metal and accessories, then the feet; after them come the colours together, formality and season. A row of the groups at the top shows each as passed, to note or failed; click one to go to it.
- Cloth. Every piece has a cloth as well as a colour: worsted, flannel, hopsack, tweed, linen, fresco, cotton twill, corduroy, denim for tailoring; poplin, oxford, twill, linen, brushed cotton for shirts; merino, lambswool, cashmere for knits; grenadine, silk twill, knitted silk, wool for ties; calf and suede for shoes; and so on, as the Style Guide lists them. Each cloth carries how formal it is, its season, and whether it shines. A piece is as formal as the lower of its type and its cloth: a tweed jacket is country wear, not business.
- 4. Sheen (Outfit Check, question 4). Anything shiny near the face fails: satin ties, lustrous worsted, glossy nylon. Below the belt it is a caution; patent shoes belong to black tie.
- Formality and the occasion. Choose the occasion at the top: casual, smart casual, business or evening. Pieces are placed on a five-step scale from casual to business, and each occasion has its range. Pieces four steps apart look like a mistake; three apart is a deliberate mix at best. The Style Guide's own pairings are checked by name: chunky trainers fail with anything smart; a tie with trainers fails; a suit with trainers is a fashion look rather than a classic one; a tie with denim needs a rough tie and jacket. A jacket and trousers in one colour but two cloths look like a suit jacket worn with the wrong trousers.
- Season. Cloth against the season switch: flannel, tweed and heavy knits belong to winter, linen, fresco, seersucker and piqué to summer, and summer and winter cloth in one outfit look as if one of them was a mistake.
- Below the belt. Tier 4 is allowed as the one bold move. Shoes are the trousers' depth or darker; light warm shoes clash with a cool top.
- Socks match the trousers or go darker, so the leg line runs unbroken into the shoe; a deep colour (burgundy, forest) is the one small play allowed there. No white socks with shoes (Style Guide, section 5).
- Pocket square. It needs a jacket with a breast pocket. White linen is never wrong; a square in the tie's exact cloth looks bought as a set.
- Eye area (Style Guide, section 1). With a marked shadow under the eyes, an outfit that is dark everywhere near the face is flagged: keep one light element there. Knitwear at the collar sits 15 to 20 cm below the eyes, too far to darken them; glasses are what sits at that height, so the You tab suggests a bottom-rim test.
Accessories
B Metal (Outfit Check, question 6): one metal through the outfit, a two-tone watch the one exception. A warm metal against cool colouring is a stray warm note. A ring worn every day is let stand outside the matching rule. Glasses and an earring are at the face, so they are compared with each other rather than with the belt. Stones are small accents: they look chosen when they echo a colour worn (lapis with navy, garnet with a burgundy tie).
B Size, each with its reason:
- Watch: 36 to 40 mm by default, or a range from your wrist (from 34–38 mm under 15.5 cm up to 40–44 mm over 18.5 cm). A larger case overhangs the wrist and will not go under a shirt cuff.
- Cufflinks: about 12 to 18 mm, showing a little below the jacket sleeve. Only with a shirt with double cuffs.
- Tie bar: about three quarters of the tie's width, so up to 60 mm on an 8 cm tie. Longer than the tie is wide, its ends stick out.
- Ring: a signet or band face up to about 15 mm.
- Earring: a stud up to about 5 mm. The pupil is 2 to 8 mm and the iris about 12 mm; a stud of pupil size reads as a point of light beside the face and the eyes stay the focus. Larger, it competes with them.
Over the range, the check lists what elsewhere in the outfit carries the piece, for example a large stud with a matte or dark face, metal matching the frames and no pattern at the face, and each one in place lifts it. Far over the range (a stud past 8 mm, a tie bar wider than the tie), nothing compensates and the check says to take it off for this outfit.
Keep this, fix the rest
For a piece you own that does not fit the outfit. Valet locks it and searches around it: first the best colour for every other open piece, then layers, putting on a jacket or a knit, a collared shirt in place of a collarless one, or taking off an optional piece, kept only when the outfit improves. If the kept piece still fails afterwards, no change elsewhere can mend it. When that failure is its colour at the face, it is not for you near the face, though the colour can still be worn lower down; otherwise it is not for this outfit.
Every row on the Palette and Outfit check tabs also says when its piece is part of a problem, with the reason, using the same judgement as the marks below. A change to one piece can make another the problem, and that row then says so.
The outfit card. Give the outfit a name and Print or save as PDF makes one page of it: the pieces from head to feet with colour, cloth, pattern and tier, the accessories, the lightness chart, and each of the card's questions with the reason for anything flagged. The name becomes the title and the PDF's file name. Every colour carries its name, because printed colour drifts.
Verdict. Following the Outfit Check card: everything passing is a good outfit; one failure with its fix applied is still a good outfit; two or more means start again from the first failure.
Clashes with your skin come first. A piece near the face in tier 3 or 4 is a different kind of failure from the others: no change to the rest of the outfit can mend it, because the problem is between that colour and your face. So whenever one is worn, the verdict says "Clashes with your skin", names the piece and gives the reason in plain words (too warm for your colouring, too close to your skin, more vivid than your colouring), taken from the colour rules of sections 4 to 7. The warning stays as long as the piece does, including when you keep or lock it, and offers the ways out: a colour that suits you, the same colour below the belt, or the full list. The lightness chart rings the piece in red and puts the clash first, on every tab. A pocket square in such a colour is noted but not failed: it is a small accent.
Suggestions and Choose the best colours. For an open piece, every colour in the library is tried in its place and the whole outfit re-checked. They are ranked by how much they fail, weighted by the rule's importance, plus the usual role of the slot: light shirts, darker ties and jackets, neutral trousers, shoes in leather colours. Those roles only break ties between outfits that pass the same rules. Choose the best colours repeatedly makes the single change that most improves the outfit, never touching a locked piece, and stops when nothing helps. It finds a good outfit, not the only one.
What still works here
B A tier says whether a colour suits you. The next choice also depends on what you already have on, and on where the new piece sits: navy is tier 1 for most men and still wrong as trousers under a navy jacket of a slightly different shade. So the library in the Outfit editor marks each colour for the piece you are choosing, and the palette on the Palette tab marks each colour for the selected place of the study (near the face, an accent, below the belt). Valet tries every colour in that place, checks the whole outfit, or the study, with the rules above, and marks it:
- Struck through: it would make a rule fail. At the face that includes every tier 3 and 4 colour; below the belt it does not.
- "!": it would bend a rule, for example a second deliberate colour.
- Unmarked: it keeps every rule as well as any colour could.
A colour is blamed only for what it causes. If two other pieces already clash, no choice for this garment can mend that, so nothing is struck out on that account and the problem is named separately. The marks change when you select another piece or place, because the rules differ by place: a shirt has to make a light-dark step with the jacket, a tie has to stand clear of the shirt, trousers have to separate from the top, shoes should be no lighter than the trousers. Nothing is hidden: seeing what is ruled out, and reading why, is how the rules become your own.
11. Patterns
A pattern is not one thing at every distance. Up close a birdseye is a field of dots; across a room it is plain navy. Valet works out what is left of a cloth at 1 m (talking), 3 m (across a room) and 10 m, and the Outfit editor shows it under "From a distance".
How big a pattern looks
A What reaches the eye is not the size of a pattern but the angle it fills: a repeat of p centimetres seen from d metres spans 2·atan(p / 2d). One centimetre fills 34 minutes of arc at 1 m, 11 at 3 m and 3.4 at 10 m. Vision science counts detail in cycles per degree: how many light-dark repeats fit into one degree. The same 1 cm check is 1.7 c/deg at 1 m and 17 at 10 m.
What the eye can see
A How much contrast the eye needs to see a fine stripe depends on its size. Campbell and Robson (1968) measured this as the contrast sensitivity function. Sensitivity peaks at a few cycles per degree and falls steeply for finer detail, reaching the limit of sight at around 40 to 50 c/deg in good light. Valet uses Barten's model of it (1999) with the parameters of ITU-R report BT.2246, checked against the colour-science library's implementation. Sensitivity also falls in dim light, and dark cloth sends less light: in a 300 lux room the limit is about 49 c/deg for white cloth and 39 for very dark cloth.
A Every cloth is a sum of stripes. Any repeating pattern can be split into simple sine-wave gratings (its Fourier components), and the eye judges each roughly on its own. Valet measured the components of each of its fourteen drawings. A component is visible when its contrast is above the eye's threshold at its size. A thin line or a small dot is made of many weak components that all peak at the line, so Valet adds the visible ones together; judging them one at a time would call a sharp windowpane line faint.
A Above threshold, contrast looks like contrast. Once a stripe is clearly visible, its apparent contrast hardly depends on its size (Georgeson and Sullivan 1975, "contrast constancy"). So Valet takes the strength of what is visible to be its contrast less the threshold.
A B Colour detail goes first. Red-green and blue-yellow differences without a light-dark difference reach their limit at about 11 to 12 c/deg (Mullen 1985), roughly a quarter of the limit for light and dark. A burgundy dot on navy keeps its colour only when it is large enough. Valet models this with a smooth factor that halves colour differences at 5 c/deg and removes nearly all of them by 12; the shape of that curve is an approximation.
Plain, texture or pattern
B The science says what is visible. Calling it texture or pattern is a convention, and these are Valet's working values:
- Plain: nothing is visible.
- Texture: something is visible, but only finer than 10 c/deg, so each light or dark element is under about 3 minutes of arc, three times the finest detail a sharp eye resolves (grain, not shapes). Or it is visible only faintly: under 5% contrast and under about 5 ΔE of colour.
- Pattern: the rest.
A colour difference counts as visible from about ΔE 2.3, roughly the smallest difference most people notice between flat colours.
Across a room decides. Whether a cloth counts towards the pattern limits at the face is judged at 3 m, the distance the Style Guide says matters. That turns several old rules of thumb into one measurement:
- A navy birdseye or nailhead reads as plain or texture from 3 m, so it is a foundation, not a pattern. Even with a white dot it is only texture there: the dots are too small to read as shapes.
- A tonal herringbone fades to texture between 1 m and 3 m; a contrasty one does not.
- A 6 mm gingham is still a pattern across a room, in navy and white or in white and pale blue. Halve it to 3 mm and both read as texture there.
- A 7 mm pin dot sits at the edge: its dots are about 5 mm apart, right at the 10 c/deg line from 3 m. A wider-spaced dot reads as pattern. The Style Guide counts pin-dot ties as a pattern; Valet's measurement says they are borderline.
B Tonal patterns, whose two colours are within ΔE2000 12 of each other (grey on grey), count as texture even where they are visible: the Style Guide's convention, kept as a separate rule.
Limits. The model describes an attentive viewer looking straight at flat cloth in a lit room (300 lux, a 40 cm field). In life people glance, cloth folds, and texture in the weave masks fine pattern, so the model tends to call visible things that would go unnoticed. Read "pattern" near a boundary as "noticeable if you look". The fourteen drawings are simplified, and real yarns are never two flat colours.
The rules for combining patterns
A From a distance a pattern becomes its average colour. For every colour rule a patterned cloth is scored as its two colours mixed by area, in linear light. A navy cloth with a fine white stripe reads lighter than navy.
B Scale is one repeat, in centimetres, checked by hand:
- Under 1 cm: semi-solid.
- 1 to 3 cm: small, about one or two finger widths. Shirts and ties.
- 3 to 9 cm: medium, up to a palm. Jackets; and the only medium pattern in an outfit.
- 9 to 15 cm: large. Coats.
- Over 15 cm: too large to wear.
B Which pattern on which garment. Valet offers only the cloths the Style Guide uses on each garment type, and flags a scale above what suits it (one class above is a bold move, two is wrong):
- Headwear: a flat cap in herringbone, houndstooth or glen check; beanies, fedoras and panamas plain. A brimmed hat also has a band and a brim: the band sits at the face, so its colour is judged there (a navy band helps a natural straw panama); the brim should follow the frame, narrow looking small on a tall or broad man and wide swamping a slight one. A hat sits against the face (on a bald head, against the scalp), so it is scored at the face like a collar and should be tier 1 or 2. It adds a dark line above the face but does not frame it; the frame is at the neck.
- Shirt: pinstripe, bengal, candy, gingham, tattersall; small scale. Roll necks, polos and knits: plain, or a Breton stripe on a crew neck or T-shirt.
- Tie: repp stripe, pin dot, foulard, paisley, small houndstooth. Bow tie: pin dot, foulard, paisley, small houndstooth. Cravat (ascot), worn inside an open collar: paisley, foulard, pin dot. All small scale, and all should be darker than the shirt. Scarf: paisley, herringbone, houndstooth, up to medium scale.
- Jacket and trousers: nailhead, birdseye, herringbone, pin and chalk stripe, houndstooth, glen check, windowpane; up to medium scale. Overshirt: also gingham. Coat: up to large scale.
- Socks: birdseye, pin dot, Breton stripe, herringbone. Pocket square: pin dot, foulard, paisley, houndstooth.
- Paisley is a figured pattern: the teardrop (boteh) with a curled tail. Foulard is a small repeated geometric figure, traditionally printed on silk. Both count as patterns like any other: scale, contrast and the two-pattern limit apply. A paisley in warm colours (gold, rust) can come to the face only inside the warm theme on a cool dark base (Style Guide, section 8a).
B Near the face, at most two patterns, with at least one plain layer, and the two sharing a colour. The usual menswear limit is three; two is the Style Guide's deliberate choice for a bald head.
A B Different scales. Two patterns of similar size interfere: the eye cannot separate them, and close stripes can shimmer (moiré). Vision science explains why similar spatial frequencies mask each other (Campbell and Robson 1968). The working rule that the larger repeat should be about twice the smaller is tailoring convention.
B Pattern contrast follows the face. A pattern near the face whose own light-dark contrast is above the upper bound of section 4 is louder than the face it frames.
On a video call
A B A camera records the world as a grid of pixels, and detail finer than two pixels per cycle cannot be recorded: it turns into false, moving stripes (aliasing, seen as moiré). A webcam framing your head and shoulders spreads about 84 cm over 1280 pixels at 720p, about 0.66 mm a pixel. Valet warns when a cloth has strong detail with a period under four pixels (2.6 mm). That is the sampling science; the four-pixel margin and the strength cut-off are working values. Fine stripes, small checks, herringbone and birdseye in strong contrast are the usual culprits, which matches the advice given to people appearing on television.
Do vertical stripes slim?
A Valet does not score stripe direction, because the evidence runs against the popular rule. In 1867 Helmholtz observed that a square filled with horizontal lines looks taller and narrower than the same square filled with vertical ones. Thompson and Mikellidou (2011) measured it: a rectangle of vertical stripes had to be made 7.1% taller to match the height of a square of horizontal stripes, and a rectangle of horizontal stripes 4.5% wider to match the width of a square of vertical ones. The effect held on a drawn figure and on a photographed mannequin. Ashida, Kuraguchi and Miyoshi (2013) found it stronger on slim figures, possibly reversed on heavy ones, and varying a lot between observers. The honest summary: horizontal stripes do not widen a slim man, and direction is a small effect next to contrast and scale.
In the chart
A patterned piece sits at the colour it adds up to, with its two colours marked as ticks, so you can see both how light it reads and how much light-dark contrast it carries inside.
12. The lightness chart
Beside your profile, the palette and the outfit sits a chart of lightness: the one quantity the eye reads first, and the one most outfit rules are about.
A What it shows. One row for you and one for each piece, head to toe. Everything sits on a scale of lightness (CIELAB L*), dark on the left and light on the right. Your skin, hair or brows and eyes are the circles on the top row; a dashed line runs down from your skin through every row. A patterned piece sits at the colour it adds up to from a distance (section 11), with its two colours as thin ticks.
A B The green zone, on the rows at the face, is the lightness that frames your face: the band of section 4, measured from your skin. A piece in it gives the face a clear edge; a piece near the dashed line separates from your face only by colour.
B The checks underneath are three of the outfit rules (section 10), with their numbers: the strongest light-dark step among the layers at the face, the step between the top and the trousers, and shoes against trousers. They use the same thresholds as the Outfit check, so the two always agree. The T1 to T4 tags are each face piece's tier.
Hover a colour anywhere to try it: the piece moves on the chart and the checks update. On the Palette tab the rows are the study's places (near the face, accent, below the belt) rather than pieces.
The Palette tab and its study
The Palette tab is about you and colour; the Outfit check is about dressing. The palette sorts every named colour by where it can go on you. Beside it, the study holds up to three colours by where they sit: near the face (a shirt, knit or jacket), an accent (a tie, scarf or pocket square) and below the belt (trousers). Choose a place and the palette is marked against what is already in the study, so you can see, for example, that forest near the face rules out forest or navy below the belt and leaves grey. The rules are the outfit rules of section 10, without the ones about layers and cloth. The study never changes your outfit; Send to the Outfit check puts its colours on the matching pieces there.
Above the palette, your core palette picks the best of each tier for the season: base neutrals, colours for the face, deliberate colours, colours for below the belt, and the colours that echo your eyes and hair, spread so that no two look alike. It is rebuilt from your profile every time, so playing in the study never changes it.
A tier says how a colour suits you; it says nothing about which colours go with each other. What goes together, under the core palette, answers that for the core colours: choose one for near the face, and every other core colour is tried with it as an accent and below the belt, by the study's rules, and sorted into goes with it, with care, and not with it (mauve takes a burgundy accent, a neighbour on the colour wheel, but not marine, which is neither a neighbour nor an opposite, so the two compete). Below it, a few tested combinations of three, each passing every rule, half with a neutral near the face and half with a colour.
Your palette is what suits you; the study is a combination drawn from it, the colours of one outfit. A combination you like can be named and saved under the study. Saved combinations are kept with your profile and in its file, and each is judged again whenever you open the list, against your colouring and the season you are in, by the same rules as the study: ✓ no rule broken, ! something worth knowing, ✕ a rule fails. Clicking one puts it back in the study, so it can be changed, sent to the Outfit check or compared with the next on the chart. The printed palette card keeps only what works: the saved, tested and current combinations that pass every rule this season (a saved one that fails is left out and counted), and for each core colour near the face, the core colours that go with it. Under Build around one colour, tick Put [colour] on the palette card too to add that colour: where it can go on you (near the face, or below the belt only, with its versions that suit the face), and, with it in that place, which core colours go in the other places, a combination that works and what to keep apart from it. Clear the study empties all three places, so the palette is marked only where a colour clashes with your skin.
Why this colour scores the colour you point at, or the one in the selected place, at your face rule by rule. The study together lists the rules between the study's colours. Print or save as PDF makes a palette card: your colouring, the four tiers with every colour named, rules of thumb worked out from your numbers, the daylight tests, and the study if one is open. Printed colours are approximate, so the card names every colour.
Colour. The chart sits on white, in both light and dark themes, so its text and rules read clearly. A surround changes how every colour in front of it looks (section 6): against white, colours look a little darker and richer than they will against a grey day or a dark room. The chart is for lightness and steps, so judge the colour itself on cloth.
Measurements. The guide on the You tab draws where the tape goes, on the body and, where possible, on a garment that fits. Each value is checked against a broad range for your height (for example, chest 42–78% of height) to catch a slipped tape or inches typed as centimetres. These ranges are deliberately wide and say nothing about how you should be built.
What it is not. A screen is not cloth. If a garment passes a daylight test at the jaw and fails here, trust the daylight.
12a. Fit and size
The Fit & size tab is switched off for now, while Valet concentrates on colour and pattern; this describes it for when it returns.
The Fit & size tab turns the measurements from the You tab into sizes, checks a garment against your body before you buy it, and lists what to look for at a try-on. Everything is a starting point: brands cut differently, and a garment's own measurements beat any label.
- A German and Italian jacket sizes are half the chest in centimetres (100 cm is size 50). German long sizes, for tall slim men, are numbered 2 × size − 2 (52 becomes 102); short sizes half the size (26).
- B UK and US jacket sizes are the chest in inches; the letter (Short, Regular, Long) follows height. Some brands advise an inch smaller for a slimmer fit.
- B Shirt collar: the neck plus about half an inch (1.25 cm), so one finger fits.
- B Drop: suits sold as a set pair the jacket with trousers a fixed amount smaller, 6 inches as standard; a slimmer cut is drop 7 or 8. When your chest-to-waist difference is larger, a standard suit gapes at the trouser waist.
- B Ease and cut. A garment is larger than the body by its ease. Tailors add roughly 10 to 18 cm to the chest of a jacket depending on the cut. Valet reads the cut a garment will have on you from its ease: for a jacket's chest, 6 to 10 cm is slim, 10 to 14 modern (tailored), 14 to 19 classic; less is tight, more hangs. Shirts and trouser seats have their own ranges. The same jacket is slim on a broad man and classic on a narrow one.
- B What a tailor can and cannot do. Each finding says whether it can be altered: sleeves and trouser hems easily; waists within the cloth folded inside, often 2 to 4 cm; shoulders, collars and jacket length hardly or not at all. A wrong size in those is the wrong size.
- B Lengths: a jacket sleeve ends just before the wrist bone with 1 to 1.5 cm of shirt cuff showing; a jacket is about half your back-neck-to-floor height long; trousers about 3 cm shorter than your inseam to the floor rest on the shoe in a slight break.
- A Hats are sized by head circumference in centimetres; US sizes are that in inches divided by π. Shoes: European sizes count the last in Paris points of two thirds of a centimetre, the last being about 1.5 cm longer than the foot. Lasts differ, so shoes are always a try-on.
13. What about "seasons"?
C Seasonal colour analysis (spring, summer, autumn, winter, and their sub-types) is popular and often gives sensible advice, because it bundles contrast, depth, temperature and chroma into named boxes. But the boxes themselves have not been validated, and people at the edges get forced into the wrong one. Valet keeps the four dimensions continuous instead. The temperature rule in section 7 is the part of Valet closest to this tradition, which is why it is badged B, carries a confidence label and never decides a tier alone.
14. Privacy
Everything happens in your browser. The photo is never uploaded or stored; only the sampled colours, your measurements, settings and the outfit (colours, cloth and locks) are kept in this browser's local storage. “Save to a file” writes them to a JSON file you control, together with your palette for winter and summer (each colour's tier, and the reading it came from), so the file can be read on its own. When a file is opened, the palette is rebuilt from its colouring with the current rules, and any colour that has changed tier since the file was saved is listed. A file that holds only a palette takes its colouring from it. Valet loads nothing from the network: every script is a file in this folder, and the page is set so it cannot contact other sites. The links to sources open in a new tab; the sites they lead to are not part of Valet and run their own scripts and advertising. No cookies, no analytics.
15. What the menswear writers say
Valet's rules about your own colouring were not invented here. They come from the literature on classic men's dress, which states them in words and shows them on examples. This section lists what those writers agree on, where each claim sits in Valet, and what Valet adds: a number, so the rule can be applied to a face that is not in the book. All of it is convention B unless a row says otherwise.
| What the writers say | Who | In Valet |
|---|---|---|
| Match the light-dark contrast of the clothes at your face to the contrast between your hair and skin. Men fall into high, medium and low contrast. | Flusser 2002; Schlueter 2019; Barker 2020; Permanent Style 2019 | Sections 3 and 4. The three groups become a measured step with bands at 25 and 45. |
| Hair counts most, then skin, then eyes. Light eyes lower the contrast of a fair-haired man and raise it for a dark-haired one. | Permanent Style 2019 | Hair's step counts in full, brows at 80%, the iris at 60%, so the eyes can add to your contrast but rarely set it. Valet does not yet let light eyes raise the contrast of a dark-haired face. |
| Grey hair changes your group: low contrast on fair skin, high on dark skin. | Schlueter 2019; Barker 2020 | The grey slider (section 3). The direction falls out of the arithmetic instead of being a separate rule. |
| Dark skin carries strong contrast; very fair skin is overpowered by black with white. | Schneider 2018; Permanent Style 2019 | The whites-of-the-eyes term and the upper bound in section 4. |
| Repeat one of your own colours, eyes, hair or skin, in the outfit. | Flusser 2002; Schlueter 2019 | "Echo" on the You tab: the library colours nearest your eyes and hair, each with its tier. |
| Skin has an undertone, warm or cool. Warm suits earth colours and gold; cool suits blues and silver; neutral can wear both. | Schneider 2018 | Section 7, including the frames rule. Kept at B and marked low confidence when judged by eye; the one experiment supports it only in part. |
| Richer, deeper colouring carries more pastels and bright white; lighter colouring suits earth tones more. | Barker 2020 | The summer profile (section 3): a tan moves white and pale colours up a tier, because they now separate from the skin. |
| Semi-solids such as birdseye and nailhead read as solid from a couple of metres; small patterns on a jacket read as texture when the contrast is moderate. | Style Guide (this project), drawing on the menswear consensus | Section 11: measured as what remains visible at 3 m. Contrast and darkness of the cloth matter as much as size. |
| A colour that does not suit you can still be worn away from the face: trousers, socks, a pocket square. | Schneider 2018 | Tiers 3 and 4 (section 8). |
| Some colours flatter everyone. The lists differ: aubergine, true red and green in one; dusky rose, true red, teal and deep purple in another. | Schneider 2018; Barker 2020 | Not assumed. Each is scored against your profile like any other colour, so you can see whether the claim holds for you. |
How much weight to give all this. The same writers warn against overdoing it. Permanent Style argues that complexion matters less than most men think: most men sit in the middle group, men wear few large areas of strong colour, and a shirt, tie and jacket together can balance almost anything. Gentleman's Gazette tells readers to wear the colours they like and move a difficult one further from the face. Valet's tiers should be read the same way, as tendencies with reasons attached. A tier 2 colour you love, worn knowingly, beats a tier 1 colour you do not.
Where Valet departs from them. The writers judge contrast by eye and by type. Valet measures it, which forces choices the books never had to make: where the bands fall, how much the eyes count, what a half-grey head adds up to. Those choices are listed in section 3 as working values. If the daylight tests on the You tab disagree with a verdict, the daylight is right.
16. Sources
Perception and colour science
- M. E. Chevreul, De la loi du contraste simultané des couleurs, Paris, 1839.
- J. von Kries, chromatic adaptation, 1902.
- A. H. Munsell, A Color Notation, Boston, 1905.
- CIE, colorimetry: CIE 1976 L*a*b* colour space.
- F. W. Campbell, J. G. Robson, "Application of Fourier analysis to the visibility of gratings", Journal of Physiology, 1968.
- M. A. Georgeson, G. D. Sullivan, "Contrast constancy: deblurring in human vision by spatial frequency channels", Journal of Physiology 252, 627–656, 1975.
- K. T. Mullen, "The contrast sensitivity of human colour vision to red-green and blue-yellow chromatic gratings", Journal of Physiology 359, 381–400, 1985.
- P. G. J. Barten, Contrast Sensitivity of the Human Eye and Its Effects on Image Quality, SPIE Press, 1999. Parameters as in ITU-R Report BT.2246 and the colour-science library (colour.contrast).
- H. von Helmholtz, Handbuch der physiologischen Optik, Leipzig, 1867.
- P. Thompson, K. Mikellidou, "Applying the Helmholtz illusion to fashion: horizontal stripes won't make you look fatter", i-Perception 2(1), 69–76, 2011. doi:10.1068/i0405
- H. Ashida, K. Kuraguchi, K. Miyoshi, "Helmholtz illusion makes you look fit only when you are already fit, but not for everyone", i-Perception 4(5), 347–351, 2013. doi:10.1068/i0595rep
- K. M. Lam, Metamerism and colour constancy, PhD thesis, University of Bradford, 1985 (the Bradford transform).
- G. Sharma, W. Wu, E. N. Dalal, "The CIEDE2000 color-difference formula: implementation notes, supplementary test data, and mathematical observations", Color Research & Application, 2005.
- R. Russell, "A sex difference in facial contrast and its exaggeration by cosmetics", Perception, 2009.
- B. Ottosson, "A perceptual color space for image processing" (OKLab), 2020.
- L. Jiang, H. Wang, C. Gao, X. Zhang, K. Xiao, M. Melgosa, C. Li, "Skin color measurements before and after two weeks of sun exposure", Vision Research 192, 107976, 2022. eprints.whiterose.ac.uk/187764
- A. Porcheron, E. Mauger, R. Russell, "Aspects of facial contrast decrease with age and are cues for age perception", PLOS ONE 8(3), e57985, 2013. doi:10.1371/journal.pone.0057985
- New Mexico State University Cooperative Extension, "Determining Personal Colors", Guide C-315. pubs.nmsu.edu/_c/C315
- D. I. Perrett, R. Sprengelmeyer, "Clothing aesthetics: consistent colour choices to match fair and tanned skin tones", i-Perception, 2021.
Classic menswear
- A. Flusser, Dressing the Man: Mastering the Art of Permanent Fashion, New York, 2002. The source most later writers credit for the contrast groups and for repeating your own colouring.
- S. R. Schneider, "How To Find The Right Colors For Your Skin Tone", Gentleman's Gazette, 15 October 2018. gentlemansgazette.com/how-to-find-colors-for-skintone
- P. Schlueter, "How to Use the Color Wheel to Assemble Superior Outfits", Gentleman's Gazette, 18 January 2019. gentlemansgazette.com/how-to-use-the-color-wheel-to-assemble-superior-outfits
- "Dressing for complexion and skin colour: Should you care?", Permanent Style, 27 March 2019. permanentstyle.com
- C. Barker, "How to Dress for Your Skin Tone", The Rake, January 2020. therake.com/default/stories/skintone
Fit and size
- U. Morlock et al., "Plus sizes men", 3D Body Scanning Technologies conference, 2016: German men's outerwear sizes are half the chest girth. proc.3dbody.tech
- "How to Measure Chest for a Suit Jacket", Nathan Tailors: raw chest plus 4 to 7 inches (10 to 18 cm) of ease by cut. nathantailors.com
- "Ease (sewing)", Wikipedia. wikipedia.org
- "How to Choose a Proper Cloth Jacket Size", Proper Cloth. propercloth.com
- "How to measure for a shirt", T.M. Lewin: neck plus half an inch. tmlewin.co.uk
- Formal shirt size guides, Charles Tyrwhitt (collar, chest and sleeve tables). charlestyrwhitt.com
- On suit drop and waist suppression, Gentleman's Gazette. gentlemansgazette.com
- "How to Measure a Suit Jacket", Rampley & Co: 1 to 1.5 cm of shirt cuff below the sleeve. rampleyandco.com
Naming skin, hair and eyes
- T. B. Fitzpatrick, sun-reactive skin phototypes I to VI, 1975. A scale of reaction to sun, not of colour.
- A. Chardon, I. Cretois, C. Hourseau, "Skin colour typology and suntanning pathways", International Journal of Cosmetic Science 13, 1991 (the individual typology angle).
- E. Monk, "The Monk Skin Tone Scale", SocArXiv, 2023. doi:10.31235/osf.io/pdf4c
- The hair-colour level system, 1 (black) to 10 (lightest blond) with tone words, as printed on professional colour charts, for example L'Oréal Professionnel, "Hair colour numbering system". lorealprofessionnel.co.uk
- The Fischer–Saller scale (hair) and the Martin–Schultz scale (eyes), reference charts of twentieth-century physical anthropology, named here only as the technical counterparts of the everyday words.
Later modules will add the tailoring literature on proportion and ease.