Glasses Frame Size Numbers Explained: Two Tests the Stamp Settles and Two It Cannot

Which pupillary distances each stamped size is inside

How the windows are built, since this is the one column set a reader can check: decentration per eye is (frame PD − your PD) ÷ 2, so a limit of L millimetres per eye opens a window from frame PD − 2L to frame PD + 2L. That conversion is this page's own arithmetic; the three limits are published figures, and this page does not claim to know which one a given laboratory honours. Sensitivity: one millimetre of lens width moves both edges of every window by one millimetre and moves the decentration by 0.5 mm (0.02 in), and the two published limits sit 1 to 2 mm apart per eye, which is 4 mm (0.16 in) of pupillary distance at the edges of every row — the 54-20-140 front is inside the 5 mm ceiling somewhere between 64 and 84 mm and inside the 3 mm reading only between 68 and 80 mm. Reachability: the adult band a hospital health article published on 17 April 2026 runs from 50 to 70 mm, so a window that extends past 70 mm is not a size you can wear at that end, and the strict columns are where that bites — of the 12 mm opened by the 3 mm reading, the 54-18-140 row keeps 4 mm inside that band, the 54-20-140 row keeps 2 mm and the 56-17-145 row keeps 3 mm. The temple length differs across rows one, five, seven and ten (135, 145, 145, 145 mm) and no column moves, because the third number is not an operand in any calculation on this page. Rows four and six are the instructive pair: 51-19-140 and 52-18-140 have the same frame PD, so their windows are identical, and they part company only at the blank — see the depth table below. Inches throughout are the millimetre figures divided by 25.4.

The dictionary, and the one thing the stamp gives you for nothing

The three numbers are stamped in a fixed order — lens width, bridge width, temple length — written either with dashes or with a small square between the first two, so 54-20-140 and 54□20-140 name the same front. The ranges they are drawn from were published on 18 January 2019 by a consumer eye-health reference and are still published with the same figures on its most recent update, in April 2026: adult lens widths run 40 to 60 mm (1.57 to 2.36 in), bridges 14 to 24 mm (0.55 to 0.94 in), temples 120 to 150 mm (4.72 to 5.91 in). That page never mentions pupillary distance. It is a dictionary of the stamp, and it is the oldest dated position in the table at the end of this page by four years and eight months — the next one, an optical retailer's definition page from October 2023, names the frame PD against your own and prints no millimetre value and no sum. The blank step appears in no consumer page read for this article at all.

One thing comes out of the stamp without asking for anything else. The effective diameter of a finished lens is defined by a laboratory reference published on 10 June 2026 as twice the longest radius from the lens's geometric centre to its farthest edge, and it is never smaller than the lens width: a lens 52 mm across has an effective diameter of at least 52 mm, and a rectangular lens is larger than that. Put that floor through the minimum-blank rule published on 5 November 2025 — effective diameter, plus the difference between frame PD and your PD, plus 2 mm — and a 52□18 front on a 63 mm (2.48 in) face needs a block of at least 61 mm (2.40 in) before any decision about depth or shape is taken. That is a real answer from the arm alone, and it is the only one: the same 52□18 front needs a block past the deepest one in that published list once the lens is 42 mm deep, which is section four.

The third number is not an optical operand at all. Temple length is where the bend sits behind your ear; it enters the decentration sum nowhere, and it enters the blank sum nowhere, which is why four rows of the table above carry 135 and 145 mm temples and every column reads as if they did not. Its only published range on the frame is a comfort range, and it is the one of the three you can have adjusted on a metal front without changing anything about the optics.

The subtraction, and the pair of numbers that are not interchangeable

Here is the whole of the first test, and it is one line: frame PD = lens width + bridge width, and decentration per eye = (frame PD − your PD) ÷ 2. The published example is the one people keep asking about, because it is the example that appears on the frames in a shop: a 52□18 front gives 52 + 18 = 70 mm (2.76 in) of frame PD, and at a 64 mm (2.52 in) pupillary distance each lens is moved inward by 3 mm (0.12 in). That subtraction and that halving are on a definition page with no publication date and no review date printed on it, and it is the only consumer page read for this article that completes the arithmetic — the rest stop at naming the quantities, or at the dictionary.

Why the limit matters at all: a lens is ground with its optical centre where the front's geometry puts it, and moving it off-centre costs edge thickness on minus prescriptions and introduces prism at the point you actually look through. So the laboratory's own ceiling is a distance, not a preference — the number in print is “keep decentration under 5 mm per eye where possible,” published 9 March 2026. A frame-styling guide twenty-five days later printed a tighter band, “3–4 mm per eye when possible,” and gave an adult pupillary distance range of 54 to 74 mm (2.13 to 2.91 in) beside a frame width range of 120 to 150 mm (4.72 to 5.91 in), which it said “determines how far the lens must be decentered.”

That last sentence is the one to take apart, because it is the only published attempt at the subtraction this article is built on, and it uses the wrong pair. Total width is not frame PD. Total width is measured end piece to end piece, so it carries the whole front plus both end-piece allowances, and it is 58 to 71 mm (2.28 to 2.80 in) larger than the frame PD of the sizes in the table above — a lens width plus the two allowances. Subtract a pupillary distance from the total width instead of from the frame PD and, on a 52□18 front that measures 138 mm (5.43 in) end piece to end piece — its 70 mm frame PD plus the lens width plus a 16 mm allowance — worn by a 62 mm (2.44 in) face, you get 76 mm, against a true 4 mm (0.16 in). Seventy-six millimetres is larger than either lens in that frame and nineteen times the limit the same guide publishes on its own page, so it cannot be the thing being measured. The charitable reading is that the sentence is directional — a wide front relative to your PD leaves the lenses far to travel — and as a direction it is right; what it is not is a procedure, and the difference between the two readings is the difference between “narrower is better” and a number you can buy to.

The true subtraction is worth having exactly, because it is the one place the stamp pays. Its sensitivity is gentle and it is not optional: every millimetre of lens width moves the decentration by half a millimetre, and every millimetre of your pupillary distance moves it by half a millimetre in the other direction. A frame PD four millimetres above your PD is 2 mm (0.08 in) of travel per eye. A frame PD ten millimetres above it is at the edge of the looser published ceiling and past the tighter one. And a frame PD below your PD gives a negative answer, which is a different failure: the centres have to travel outward, which the 5 mm rule still measures but which the blank rule of section four was not written for.

The two tests the stamp cannot answer

Two of the four things a frame has to satisfy are not in the stamp at all. Depth — the vertical height of the lens opening, the B of the boxing system — is almost never printed, and total width is never printed, because the two millimetres of metal or acetate at each end of the front, the part that carries the hinges into your temples, vary by material and by manufacturer. Neither can be recovered from the three numbers you can read. What they cost is asymmetric, though, and the arithmetic is short enough to work in front of a shelf: with the frame PD and a pupillary distance fixed, depth is the only moving part in the blank sum, and it moves the answer from one stocked diameter to the next and then off the end of the list.

The table below fixes everything but depth. Front 52□18-140, so a frame PD of 70 mm; pupillary distance 63 mm, so the difference term in the blank rule is +7 mm for every row; effective diameter taken as the diagonal of the boxed rectangle, which is this site's own upper bound and slightly overstates a real lens, because real lenses have rounded corners. The right-hand column is the largest diameter commonly stocked — 65, 70 or 75 mm — since the published failure mode is not a bad fit but a lens that cannot be cut.

Sensitivity, in the two directions that matter. Depth: across this column each extra millimetre adds 0.45 mm at the shallow end and 0.63 mm at the deep end to the effective diameter, because the square root flattens — and the ceiling that falls out of the last stocked size is a depth of 40.64 mm (1.60 in) for this front at this pupillary distance, 31.89 mm (1.26 in) if the shop's largest block is 70 mm, 20.78 mm (0.82 in) at 65 mm. Pupillary distance: the difference term is linear, so a 71 mm (2.80 in) PD instead of 63 mm drops every requirement in column three by 8 mm and puts the 42 mm lens back inside a 70 mm block, which is the reason this site's round-face glasses page warns that the blank rule subtracts when your PD is wider than the frame's and that it was written for the inward case. The pair from the first table settles the division of labour between the two tests: 51-19-140 and 52-18-140 have the same frame PD and therefore identical pupillary-distance windows, but at a 33 mm (1.30 in) depth the first needs 69.75 mm (2.75 in) and the second 70.59 mm (2.78 in) — one millimetre of lens width, two different stocked blanks, and nothing on the arm that tells you which lens you are holding until you ask for the depth.

Two ceilings in print, and which window to buy to

The three right-hand columns of the first table are not three readings of one rule; they are the state of the published advice, which is that nobody has settled the number. The ceiling in the trade literature is 5 mm per eye where the frame allows it. The tighter band, 3 to 4 mm per eye when possible, appeared in a consumer-facing guide in April 2026 with no citation and no derivation, and it is the one a person ordering high-index lenses for a strong prescription will care about. The distance between them is 4 mm (0.16 in) of pupillary distance at each edge of every window, which is not small: for the 50-16-140 front it is the difference between a 20 mm band of legal faces and a 12 mm one.

Buy to the narrow window if any of three things is true: your prescription is minus and past about −4.00 dioptres, where edge thickness at the frame's outside corner is what people notice; you are ordering online and cannot have the front mechanically re-centred afterwards; or your PD sits near an edge rather than in the middle. Buy to the 5 mm ceiling if your PD is comfortably inside it and the frame is a small one, where the arithmetic stops being the binding constraint and total width is. That is a judgement about which constraint binds, not a third published limit, and this page labels it as its own.

The case where the windows stop helping is the outward one. When your pupillary distance is wider than the frame's PD, the subtraction goes negative, and the decentration is measured from the centre outward: the 46-15-135 front, at 61 mm of frame PD, needs 1 mm of outward travel on a 63 mm face and 5 mm (0.20 in) on a 71 mm (2.80 in) one. Outward travel is not illegal under either published ceiling, but it is the direction in which the blank rule's subtraction starts taking material off a number that is already only the lens's own diagonal, which is why the same rule appears here in its inward form and why this page does not treat a negative answer as extra headroom.

• Check the arithmetic before the aesthetics: twenty-nine rows across three tables are on this page, and the one you can do in your head is lens width plus bridge width.
• Find your own number on paper first — a prescription carries it, and a shop that made you glasses before has it on file.
• Write your PD beside the frame PD of anything you like. If the gap is 10 mm or less, the looser published ceiling is satisfied; if it is 6 mm or less, both are.
• Ask for the depth of any frame you are close to buying, with a ruler, vertically at its tallest. It is the number that decides the blank.
• Ask for the total width too, end piece to end piece, and compare it against your own cheekbone line rather than against the size on the arm.
• If your PD is wider than the frame's, stop and ask what the laboratory will do about outward decentration before you commit to the front.

Thirteen published positions, by the date they were published

The reason this table exists is the same reason the arithmetic above is spelled out: the pieces of one calculation are published in different places, by different kinds of publisher, at different dates, and no single page a reader can find puts them in order. Dates below are the day the position was published, with a page's own later update line named separately where the two differ; no links are given to any of them.

Read the last column down: thirteen rows, and each one carries exactly one of these. Three stop at the dictionary — a lens width of 40 to 60 mm, a bridge of 14 to 24 mm and a temple of 120 to 150 mm, printed on 18 January 2019 and still there after that page's own April 2026 update, which makes them the most stable numbers on this page and the ones every consumer guide in print is still content to stop at. One names the subtraction without computing it, two hand over an operand of it, and one runs it with the wrong pair of operands. One runs it to the end on a real stamped size and carries no date at all. Two carry the blank arithmetic, one carries the ceiling, one defines the effective diameter the blank needs, and one measures a remainder on somebody else's data. What moved across the seven years and seven months the table spans is the ceiling — two numbers published twenty-five days apart in 2026 that differ by 1 to 2 mm per eye, which is the only entry here that changes a verdict in the first table.

Coverage runs from 18 January 2019 to 14 September 2026 — seven years and seven months precisely, nearly eight years when a row needs an integer, and this page does not round the span up. The 2026-08-23 row is the only position in the table whose numbers moved between the day this site first cited them and the day this page was written — the same address now publishes lower per-material remainders and a larger sample, and says plainly that the figures describe retailer listings rather than measurements — so it is quoted here as a band rather than as a per-material ladder. Column two carries institution types and no names, and column three carries no links, on purpose.