Sunglasses for a Round Face: the Five Frame Rows, in Millimetres

Table A: the five frame rows across five outlines, in millimetres

• Four rows repeat across the columns because those four rows do not read the outline. Total width reads the width of a face, bridge width reads the width of a nose, temple length reads the circumference of a head, and lens width is what is left once the first three are settled. A round face and an oblong face of the same width are given the same front by every size chart checked for this page.
• The height row repeats nothing. Its ceiling is set by how much room there is between a brow line and a cheekbone, and that room is the one measurement on a face which the outline genuinely changes: 7 mm (0.28 in) between the round and the oblong column at the width this table is worked at.
• The width row is a union, not a consensus. Two schools of fitting guide give answers for it that are 6 mm apart at a 136 mm face, and this site is not able to reconcile them into one number without choosing a look. The section on the overshoot gives both readings, the ceiling both agree on, and the reason the disagreement survives.
• The height row's floor of 40 mm (1.57 in) is a stocked floor, not a fitted one. Optical fronts are cut at 30 – 34 mm (1.18 – 1.34 in) and the fashion bands begin at 40 mm; below 40 mm a tinted lens stops doing the thing a tinted lens is for, which is keeping light out from under the brow.

The one row that follows your width

• Two things fall out of that grid, and they are worth more than the numbers themselves. The first is that the ceiling climbs about 7 mm (0.28 in) across the outlines at a fixed width, and about the same 7 mm across two widths at a fixed outline. Your face width moves the vertical budget as hard as your face shape does. Someone 20 mm wider than you and shaped nothing like you will be buying the same lens height. The second is that round and square land on the same number in all three columns, which is not a rounding artefact: in this site's rule set those two outlines have identical length-to-width means, and what separates them is the jaw angle, a thing no frame parameter on the inside of a temple arm measures.

How the height ceiling was calculated, since none of the pages checked publishes one

Nothing in the ten pages that rank for this query, or in the size charts checked alongside them, prints a maximum lens height. What they print instead is a prohibition — the bottom rim must not rest on the cheek — which is a relational statement and cannot be looked up in a shop. So the ceiling here is built.

Face length, taken from the hairline to the chin, divides into three near-equal parts: hairline to brow, brow to the base of the nose, nose base to chin. The middle third is the corridor a sunglass lens has to live in. This site's published rule set defines each outline by a mean ratio of face length to cheekbone width — 1.103 for round, 1.103 for square, 1.182 for heart, 1.195 for diamond, 1.208 for oval, 1.169 for triangle, 1.261 for oblong. Multiply that ratio by your own cheekbone line, divide by three, and you have the corridor. At 136 mm the round face's corridor is 50 mm (1.97 in) and the oblong face's is 57 mm (2.24 in).

Six millimetres (0.24 in) comes off that. The reason is the brow: a rim that sits exactly on the brow line presses it, and a rim that stops exactly at the bottom of the corridor is touching the cheek already, because the corridor is measured to the base of the nose while the cheekbone apex sits above it. Six millimetres is this site's allowance, not a published constant, and it is the one input in this table a reader may want to argue with. Reduce it to 3 mm and every ceiling gains 3 mm; a frame bought on the tighter allowance fails on the face rather than in the arithmetic, which is the direction most people prefer their errors.

That leaves the numbers in Table B, and they are theoretical. They are not snapped to the heights the trade stocks: the charts that list lens heights at all print rungs at 40, 43, 46 and 50 mm, so a ceiling of 44 mm (1.73 in) buys a 43 mm lens on an order form, and the next rung up is already past the corridor. The tables keep the unsnapped figures on purpose, because which rungs exist is a shop's constraint and how tall your corridor is happens to be yours.

How far past the cheekbone: two schools and one ceiling

The published instructions for the width row disagree by an order of magnitude, and both readings are in Table A's cell.

The narrow school puts the front 0 – 4 mm (0.16 in) past the face for a narrow face and 4 – 8 mm past it for a wide one; a second guide in the same school asks for two to four millimetres past each cheekbone, which is the same 4 – 8 mm read from the middle out. The wide school asks for a centimetre to two — 10 – 20 mm past the widest line — on the grounds that the gap between rim and cheek is what makes a face read smaller. At a 136 mm cheekbone line the two schools bracket 140 – 144 mm (5.51 – 5.67 in) and 146 – 156 mm (5.75 – 6.14 in), and that union is what the first row of Table A is showing you.

Both schools agree on one thing, and it is the more useful of the two: there is a ceiling. Past roughly 30 mm (1.18 in) beyond the widest line on the face, the reading stops being about the face and starts being about the frame — a proportion problem rather than a fit problem, and the reason an oversized pair can look deliberate at 15 mm past and accidental at 35.

The disagreement survives because the two schools are not measuring the same line. A wrap front sits forward of the cheekbone at its ends, not beside it, so 20 mm of extra width in flat projection is not 20 mm of extra width in the plane where the fit is felt — a frame can be 16 mm wider on the tape and 6 mm wider on the face. That reconciliation is this site's inference, not a published finding, and it is testable: measure a curved front flat across its widest point, then measure it again in the curve as it sits on your face with the temples open, and the two numbers are not the same. On the prescription side of this family the same subtraction lands much smaller — 6 – 8 mm (0.24 – 0.31 in) — because an optical front is flatter, and it is worked through on this site's oval-face eyeglasses page.

The subtraction the guides leave half done

The stamped pair of numbers and the total width are not connected by the arithmetic the guides print. Add two lens widths to the bridge and the sum for the bands in Table A is 116 – 130 mm (4.57 – 5.12 in). The fronts those same guides sell are 140 – 156 mm across. What is missing is neither lens nor bridge: end pieces, the set of the hinges, and the panoramic curve of a sunglass front take 10 – 40 mm (0.39 – 1.57 in) of the total, which is between 6% and 29% of the width you are paying for. Worked on real numbers, a 50□22 front that measures 145 mm (5.71 in) is carrying 23 mm (0.91 in) of non-lens; a 54□17 at 145 mm carries 20 mm (0.79 in); a 56□15 at 150 mm carries 23 mm.

The instruction to allow "a few millimetres" for the frame edges is the reason so many mail-order sunglass frames finish short of the face they were bought for. On a flat optical front a few millimetres is the right order of magnitude. On a curved front it is wrong by about 15 mm (0.59 in), and the only way to catch it before the parcel arrives is to look for the total width, which almost every seller hides and which the stamped numbers cannot be made to predict.

Seven things a photograph of a stranger cannot tell you

The table gets you to a shortlist. The shortlist is decided on the face, and the seven criteria below only exist once you are wearing the thing. A screen cannot reach any of them, because all seven are dynamic or lateral: a smile, a head turned forty-five degrees, pressure that arrives after a quarter of an hour. What a virtual try-on can compare is two numbers — your cheekbone line and the seller's total width — and that is the entire argument for putting Table A above the fold rather than below the styling advice.

The rim that clears your cheek while you are sitting still arrives on it the moment you smile. This is the failure mode of a lens height that passed Table B, because the corridor shortens when the mid-face rises, and it is why the published instruction on the Rx side — the rim must not touch, even smiling — is worth more than any number in this article. To check it you have to smile properly, with the cheeks up, and watch the bottom of the lens rather than the middle of the frame. A round face is more exposed to this failure than an oblong one by exactly the 7 mm (0.28 in) the outlines are apart, which is a little over two of the rungs those charts print.

Temples that run straight back when you look at them from the front can flare where they pass the ear. The test is to turn your head forty-five degrees and look at the arm from the side, in a mirror, wearing the frame: an arm that kicks outward is not clamping, and a frame that is not clamping is being held on by its nose pads alone, which is the arrangement that slides. The published tolerance for the gap between arm and head is 2 – 4 mm (0.08 – 0.16 in), and the visible cost of flare is width — the temple region reads wider, which is precisely what a round face is buying sunglasses to avoid.

A front that is too narrow announces itself at the temple, not at the nose. Tight frames press the side of the head and produce a headache behind the eye rather than a mark on the bridge, and because the pressure is behind the ear line, it does not show in a photograph and does not appear for ten or fifteen minutes. This is the failure a 140 mm purchase on a 150 mm face produces, and the guide that tells you to size to a frame you already own reproduces it exactly if the owned frame was the wrong one.

The two errors are not symmetric, and you only learn that by making one. A front too wide can be shortened: a shop heats the ends of the temples and bends them in, and a wrap front with a little extra to work with is a frame you can tune for the rest of its life. A front too narrow has no such repair, because there is nothing to add — the end pieces and the curve are the width, and an optician will not fabricate either. Buying on the wide side of the overshoot is therefore the mistake that stays correctable inside the 30 mm ceiling, and buying short is the one that ends in a bin. The complaint that arrives with a narrow front is not usually about the front at all; it is a headache at the outer corner of the eye, and most people blame the sun, the prescription, or themselves before they blame the hinge line.

Another is not a fit problem but a lens problem, and it is the one no try-on in a shop will show you. A tinted lens can carry enough internal stress to bend a straight edge at its periphery while the frame fits perfectly, and the symptom is visual strain an hour into wear rather than anything visible on the face. The check is to sight a door frame, a tiled floor, or a horizon through the outer half of each lens and watch the line bow; the stress only becomes obvious against a straight reference at an angle, which is why it is a walk to the car park and not a look in a mirror. Frames that flex a lot under the hands tend to hold more of it once the lens is set, and nothing printed on a temple arm says which is which.

A habit does the same to the fit, and it quietly invalidates the flare check on a pair you have owned for a season. Sunglasses pushed up onto the crown spread the arms a little each time, and the spread is not even: the front ends up wider than the tape said it was, sitting differently on the pads, and the arm that now kicks outward at forty-five degrees may be reporting what the head did rather than what the shop sold. A pair that fits in its first month is therefore not evidence that the same model will fit in its ninth.

Last is the room the fitting happens in. A frame is tried on dry, at a temperature nobody stands outside in, and nose pads grip far better dry than damp; the slippage people complain about in July is invisible to a test done in a hallway mirror. That is why the ten-minute sequence below is a screen and not a verdict. It removes the frames that fail on geometry and leaves the ones that fail on mass to be settled outdoors.

Ten minutes on the face, in the order that catches the most

Wear them indoors for a full ten minutes before deciding anything, in this order. Smile hard and look at the bottom of the lens, not at the middle of the frame — the rim should still be off the cheek. Turn your head forty-five degrees to a mirror and look at the arm from the side; the gap should be 2 – 4 mm (0.08 – 0.16 in) and the arm should be running to the ear rather than kicking past it. Look down at the floor and back up, then up and back down: a front that travels more than about 2 mm (0.08 in) on the nose is one whose mass the bridge is not carrying, and that 2 mm is a threshold this site is proposing rather than one anybody publishes. Bend the lashes into the tint deliberately and let them go — if they meet the lens, the frame is sitting closer than the height row asked, and on a moulded bridge with nothing to bend there is no fix at all. Then walk to a straight edge and look through the outer third of each lens. Take the frame off at the twenty-minute mark and look at the skin above the nose rather than at the mirror: a flat bridge with moulded pads leaves a mark that shows up late and aches earlier, and a mark is a number the pads were not able to argue with. Put the frame back on and go for the third hour outside. If you find yourself pushing it up with one finger every few minutes, the temples are not carrying the mass, and on a sun frame that is more often true than it is with spectacles, because the arms are straight and stiff where an optical arm is curved to hook behind the ear — there is less geometry holding the back end down, so the whole front leans on the nose, and the leaning is what a fitting room never shows you.

None of those checks is a measurement of the face; each is a measurement of a frame, taken on a face that has decided nothing yet. Read against Table A they become pass or fail rather than an impression, and the row they protect hardest is the height row, which is the only one your outline writes.

The row that is not stamped anywhere either

Weight is a millimetre-adjacent number and it moves with lens area, which moves with the height row. A 54 × 44 mm lens presents about 1.4 times the area of the 52 × 32 mm optical lens it replaces, and the front gets heavier in proportion — with no prescription relief, since a tinted lens is thinner than the ground glass it stands in for. The published bands for frames alone run 8 – 14 g (0.28 – 0.49 oz) for titanium, 14 – 20 g (0.49 – 0.71 oz) for the moulded nylon materials, and 25 – 40 g (0.88 – 1.41 oz) for acetate, and the under-20-gram line is where most guides put the boundary of all-day comfort.

That band is why the bridge row and the weight row have to be read together. A metal front with separate pads can be tuned: the pads are bent, the gap opens, the frame sits higher. A moulded plastic bridge is one piece with the front and cannot be adjusted at all, which is why the low-bridge variants exist as a category, and why at the top of that acetate band an unadjustable front on a flat bridge will end the day on the nose tips rather than where Table B put it. Look down while you are trying one on — the frame's own weight is what moves it, and none of the charts checked here mentions the grams.

Where the prescription stops mattering, and where it starts again

Nothing in this article is a decentration limit, and for a plain tinted lens there is no reason it should be. A lens with no power in it has no optical centre to place, so the size ceiling that governs a spectacle lens — the diameter a blank can be cut from, and the arithmetic of getting two centres onto two pupils — is switched off entirely. That is the single structural difference between the table above and the millimetre table on this site's round-face eyeglasses page, which is built around that ceiling and reaches 40 mm (1.57 in) of lens depth as a limit for the same face. Sunglasses are free of it. A 60 mm (2.36 in) lens that a laboratory would refuse to cut for a −6.00 D prescription is stocked by the thousand as a tint.

Order the prescription version of any of these frames and the ceiling comes straight back, and it comes back onto the two rows you thought were settled: lens width and bridge width, because those two add up to the frame's pupillary distance. If you are buying sunglasses to be made with single-vision or progressive lenses, read the eyeglasses page first and treat Table A as a floor rather than a limit.

Which row you are actually shopping in

Measure the cheekbone line first, hair pushed back, tape across the tops of the bones and not around them. Then decide which overshoot you are buying — the 4 – 8 mm of the narrow school, which reads as a frame that fits, or the 10 – 20 mm of the wide school, which reads as a frame that flatters — and hold either of them to the 30 mm ceiling. Then take your outline and your width from Table B and refuse a lens taller than the number you land on. Only after those three steps is the shape word worth opening: it edits one row by at most 7 mm (0.28 in), and it does not edit the other four at all.